Secondary Logo

Journal Logo

Infectious Disease: Original Article

Long-term Impact of Human Papillomavirus Vaccination on Infection Rates, Cervical Abnormalities, and Cancer Incidence

Bogaards, Johannes A.a,b; Coupé, Veerle M. H.a; Xiridou, Mariab; Meijer, Chris J. L. M.c; Wallinga, Jaccob,d; Berkhof, Johannesa

Author Information
doi: 10.1097/EDE.0b013e31821d107b

Persistent infection with human papillomavirus (HPV) is the necessary cause of cervical cancer.1–4 There are at least 14 high-risk HPV types with oncogenic potential, but HPV-16 and HPV-18 stand out as the most important. These 2 types are associated with 70%–76% of the invasive cervical cancer cases worldwide.5

In large randomized trials, vaccination against types 16 and 18 has been shown to be nearly 100% effective against HPV-16/18-positive cervical intraepithelial neoplasia (CIN) grades 2 and 3, the major cervical cancer precursor lesions.6,7 The efficacy has been demonstrated in HPV-16/18-naive women, with no effect in HPV DNA-positive women.6,7 Therefore, vaccination is expected to be most effective when given to girls at an age when HPV-16/18 infection is not yet common.

HPV vaccination has recently been added to the national immunization program of many developed countries. In the Netherlands, vaccination began in 2009 with a catch-up campaign for girls born in 1993 up to 1997 (ie, girls aged 13–16 years at the start of 2010); from 2010 onward, vaccination is offered to all 12-year-old girls.8 This decision was informed by cost-effectiveness analyses that assumed a vaccine uptake of at least 85% and that ignored the benefit of herd immunity.9–12 As of 2010, the acceptance rate is much lower than anticipated: around 50% of eligible girls participated in the vaccination program. The less-than-expected vaccine uptake may have significant implications for the cost effectiveness of vaccination. The total effects of the vaccination program decrease as the vaccine coverage decreases, but this relation is nonlinear if herd immunity provides indirect cervical cancer protection to nonvaccinated women.13,14 Herd immunity arises when HPV vaccination significantly reduces the rate at which susceptible men and women contract high-risk HPV infection as a result of a reduced transmission of vaccine-preventable HPV types.

Various mathematical models have been developed to study the population impact of HPV vaccination. The indirect protective effect of vaccination on nonvaccinated women can be addressed only in dynamic models. Most analyses of HPV vaccination have not used dynamic models (refer the overview by Kim et al15); of the analyses using dynamic models, most have included only HPV-16 and HPV-18, assessing the impact of HPV vaccination on total cervical cancer incidence by extrapolation from the impact on HPV-16/18-associated cancers. Likewise, the impact of HPV vaccination on screening outcomes has been given scant attention in the context of dynamic modeling. There is potential for substantial inaccuracy in extrapolating the impact of HPV vaccination on screening outcomes and cancer incidence from the impact on cervical abnormalities and invasive cancers attributable to HPV-16 or -18, as elimination of vaccine-type HPV can unmask underlying high-risk HPV infections or lesions caused by nonvaccine types.16

We attempt to disentangle the direct and indirect (herd immunity) effects of HPV vaccination on cervical abnormalities and cancer incidence while accounting for all known high-risk HPV types. Based on a deterministic dynamic model, we have estimated type-specific parameters for HPV transmission and development of precursor lesions, using data from 2 surveys on sexual health and behavior and from one population-based screening cohort that contains longitudinal information about the infection status of 14 high-risk HPV types.17 With this dynamic model, we predict type-specific rates of HPV transmission in a partly vaccinated population and use these as input in an individual-based simulation model of cervical carcinogenesis, which explicitly accounts for multiple concomitant high-risk HPV types that invade, persist, and clear independently. The relatively high occurrence of multiple infections is modeled through variation in sexual activity.10,17 This approach circumvents the need to incorporate type-specific interactions in transmission and provides a plausible paradigm for describing concurrent HPV infections.18,19


To assess the long-term impact of HPV vaccination in the Netherlands, we assume that (1) the age distribution of the population is stable for both men and women; (2) organized cytologic screening will continue as currently provided,20 with constant participation over time; (3) vaccination is started in an endemic prevaccine situation; (4) vaccination is fully effective in preventing HPV-16/18 infection in HPV-16/18-naive girls; (5) vaccination confers lifelong prophylactic immunity against types 16 and 18; and (6) vaccination in the catch-up campaign achieves a similar coverage as among 12-year-old girls. We evaluate various levels of vaccine coverage, but most of the results we present relate to 50% coverage, the level achieved so far among eligible girls in the HPV vaccination program, and to 90% coverage, the target for all vaccines included in the Dutch national immunization program. As a conservative scenario for the long-term impact of HPV vaccination, we also consider a constant 50% vaccine uptake in combination with a relatively short duration of vaccine protection: full efficacy for the first 10 years followed by an exponential decrease with a half-life of 5 years, yielding a median duration of vaccine protection of 15 years. The longest follow-up study for a licensed cervical cancer vaccine has recently reported sustained efficacy and immunogenicity 8 years after vaccination.21

To control the computational burden of including all high-risk HPV types, we used a hybrid approach wherein the type-specific forces of high-risk HPV infection were calculated with a dynamic model (one that describes transmission of each high-risk HPV type separately) and were subsequently used as input in an individual-based model (which integrates all high-risk HPV types and simulates individual health trajectories of girls according to specific birth cohorts). Large-scale HPV-16/18 vaccination will result in reduced infection rates over time, and thus the forces of infection depend on both age and calendar time. By simulating individuals of various birth cohorts and combining the results, the individual-based model can, in addition to providing cohort-specific estimates, also provide temporal trends.

Figure 1 provides a schematic presentation of the overall structure of the model. A deterministic transmission model is used to calculate type-specific forces of high-risk HPV infection, which are stratified by birth cohort and calendar year. These forces of infection served as input to a stochastic individual-based model for cervical carcinogenesis. The natural history of disease in an individual woman is represented as a sequence of transitions between health states. States are stratified to reflect different levels of detail. For example, states reflecting high-risk HPV infection are stratified by 14 virus types, whereas states reflecting cervical intraepithelial neoplasia (CIN) are stratified by histologic grade, whether a lesion is associated with abnormal cytology, and whether a woman participates in a particular round of the cervical screening program. Progression of cervical cancer, according the FIGO (Fédération Internationale de Gynécologie et d'Obstétrique) cancer staging system, is reflected in increased symptomatic detection and reduced survival probability.

Flow diagram of model.

The high-risk HPV transmission model has been described in detail elsewhere.17 To reiterate, the model has a separate compartment for virgins and distinguishes 3 levels of sexual activity once individuals are sexually experienced: an average of fewer than 1 partner per year, between 1 and 5 partners per year, and more than 5 partners per year. Rates of switching among the sexual-activity categories were set to reproduce the age-specific proportions in the various categories reported in 2 large-scale surveys on sexual health and behavior in the Netherlands.22,23 Sexual-contact rates and age-related partner preferences were also obtained from these surveys. We estimated type-specific parameters for viral transmissibility, clearance and persistence, clinical progression, and natural immunity by fitting this model to the age-specific prevaccine prevalence of high-risk HPV infection in a Bayesian framework. Rates of cumulative clearance and progression up to CIN3 were obtained from a trial of HPV DNA testing in women with known cytologic status.24,25 Vaccinated women were explicitly included in the model (eAppendix, We calculated the risk of high-risk HPV infection in a partly vaccinated population by sampling directly from the joint posterior distribution of type-specific parameters, which provided 95% credible intervals (CIs) for the estimated impact of vaccination on infection rates.

The effect of HPV vaccination on screening outcomes and cervical cancer was calculated with the individual-based model, using the type-specific transmission parameters as input. More precisely, parameters of the transmission model were estimated by Bayesian methods, and the posterior means of type-specific parameters were used as input in the individual-based model. The individual-based simulation model of cervical carcinogenesis considers 14 high-risk HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68).10,26 For the purpose of the current analysis, the model was adjusted to include type-specific resistance to reinfection and to allow for switching among sexual-activity categories. This ensures that the individual-based model has similar structure and parameterization as the transmission model. The principal distinction between the 2 models is that the individual-based model incorporates multiple type-specific infections (each type having an independent risk of clearance or clinical progression), accommodates a detailed implementation of cytologic screening according to current Dutch guidelines,20 and specifies the progression from high-grade cervical lesions to invasive cervical cancer. The individual-based model yielded a satisfactory fit to prevaccine data on the overall prevalence of high-risk HPV infection in normal smears, the proportion of screened women with detected CIN2/3 lesions and the hazard of cervical cancer up to 65 years of age (eFig. 1,

To give a comprehensive account of the population-level impact of HPV vaccination in relation to vaccine coverage, results are presented along 3 dimensions. First, we show cohort-specific estimates of the impact of HPV vaccination on the age-specific and lifetime infection risks of vaccine-preventable types. These results were derived from the type-specific transmission model. Next, the impact of HPV vaccination on age-specific and lifetime cervical cancer risks is presented for various birth cohorts. These results were derived from the individual-based model, combining all high-risk HPV types. The cancer risk without herd immunity was calculated from a model with time-invariant forces of infection (similar to the endemic prevaccine situation), whereas the cancer risk with herd immunity was calculated from a model with time-dependent forces of infection—thus accounting for reduced infection rates over time since introduction of vaccination for vaccine-preventable types. Finally, temporal trends in infection rates, cervical cancer incidence, and screening outcomes are presented under the assumption of either lifelong or fast-waning vaccine efficacy. Those estimates were obtained by combining estimates from various birth cohorts. To incorporate parameter uncertainty of HPV vaccine types in assessing the vaccination impact under lifelong vaccine efficacy, we used the forces of infection from the transmission model that corresponded to the 2.5th and 97.5th percentiles of the lifetime infection risk of HPV-16 and HPV-18.


Impact of Vaccination on HPV-16/18 Infection Risk

Vaccination at 50% coverage has a large effect on the hazard of HPV-16/18 infection among girls of the 1997 birth cohort (age, 12 years in 2010, when vaccination starts) who do not comply with HPV vaccination (Fig. 2A, B). Girls born in 1992 (age, 17 years in 2010) are not eligible for HPV vaccination and would experience only a moderate reduction in the hazard of HPV-16/18 infection over their lifetime. Girls who are born after 1997 experience a further reduction in the incidence of HPV-16/18 infection because they reach sexual maturity when the new endemic equilibrium is established. Although the peak force of infection shifts to somewhat older ages as a result of vaccination, the incidence of HPV-16/18 infection continues to peak around age 20 (Fig. 2C, D).

The age-specific force of infection for female (A) HPV-16 and (B) HPV-18 infection, and the age-specific incidence rate of new (C) HPV-16 infections and of new (D) HPV-18 infections in women at 50% vaccine coverage. The upper solid curves denote the force of infection (or incidence rate) prior to vaccination, whereas the lower solid curves denote the force of infection (or incidence rate) with the new endemic equilibrium. Dashed and dotted curves give the force of infection (or incidence rate) for nonvaccinated women who are 12 or 17 years of age, respectively, when the vaccination campaign is started (gray lines).

A woman's mean lifetime probability of infection in the prevaccine era is estimated to be 0.69 for type 16 and 0.68 for type 18. Vaccination leads to a reduction in lifetime infection risk that becomes noticeable among persons born from 1980 onward. Persons born in 2010 (around the start of the vaccination campaign) already experience close to maximum benefit from herd immunity when they reach sexual maturity, irrespective of vaccine coverage (eFig. 2, At 50% vaccine coverage, the probability of ever becoming infected with HPV-16 for nonvaccinated women drops from a median of 0.73 (95% CI = 0.50–0.85) to 0.52 (0.32–0.68). For HPV-18, the decline is from a median of 0.70 (0.46–0.79) to 0.45 (0.26–0.57). At 90% vaccine coverage, the lifetime infection risk in nonvaccinated women ultimately drops to 0.12 for HPV-16 (0.02–0.19) and to 0.06 for HPV-18 (0–0.11).

Impact of Vaccination on Cervical Cancer Risk

The hazard of cervical cancer in the prevaccine era peaks at approximately 15 cases per 100,000 women-years between 35 and 40 years of age. Our model indicates that the hazard among vaccinated women also peaks between these ages (Fig. 3A, B). The peak hazard is reduced to 8 cases per 100,000 women-years if vaccination is started at age 16 (data not shown); vaccination at age 12 gives a further reduction to 5 cases per 100,000 women-years (Fig. 3). This difference is due to the proportion of vaccinated girls who have already been infected by HPV-16/18 at age 16. The hazard reduction among nonvaccinated women depends on both birth cohort and vaccine coverage. Nonvaccinated girls of the 1993 birth cohort do not experience a noticeable reduction in the hazard of cervical cancer at 50% coverage, as compared with the situation without vaccination (Fig. 3A). The reduction in the hazard of cervical cancer among nonvaccinated women increases with the birth cohort year. For those born from 2010 onward, the peak hazard among those not vaccinated becomes similar to the peak hazard among those vaccinated at 90% coverage (Fig. 3B). However, nonvaccinated women continue to experience a greater hazard after age 40 than those who are vaccinated.

The age-specific hazard of cervical cancer per 100,000 women-years of follow-up if HPV-16/18 vaccination achieves (A) 50% coverage or (B) 90% coverage. The upper gray curves denote the cervical cancer hazard in the prevaccine era and the lower gray curves denote the cervical cancer hazard for girls vaccinated at age 12. Curves in between apply to nonvaccinated women of various birth cohorts (see legend). Vaccination is offered at age 12 to all girls born in 1997 or thereafter, with a catch up to girls born in 1993–1996. Those vaccinated at age 12 still have a nonzero cervical cancer risk due to high-risk HPV types not included in the vaccine.

For girls vaccinated at age 12, the lifetime risk of cervical cancer drops by 69% (95% CI = 60%–74%) assuming lifelong vaccine efficacy to vaccine-type HPV. On a cohort basis, the lifetime risk of cervical cancer decreases nonlinearly with increasing vaccine coverage due to herd immunity. The relative importance of herd immunity, expressed as the percent of indirectly averted cases out of the total cervical cancers averted, monotonously decreases with increasing vaccine coverage (Table). The number of cancer cases averted among nonvaccinated women is highest between 50% and 70% coverage, approximating 70 cases per 100,000 women born from 2010 onward. At these levels of vaccine coverage, around 1 in 4 cervical cancers averted is due to herd immunity. The scope for an indirect effect of vaccination becomes restricted at higher vaccine coverage, as the majority of women are vaccinated and thus already directly protected against cervical cancer.

TABLE. Lifetime Cervical Cancer Risk by HPV Vaccine Coverage With and Without the Inclusion of Herd Immunity and Cervical Cancer Cases Averted Due to Herd Immunity

Temporal Trends Under Lifelong Vaccine Efficacy

After initiation of the vaccination campaign, the decline in high-risk HPV prevalence among young women becomes quickly noticeable, but it takes several decades before the population prevalence of high-risk HPV infection attains a new endemic equilibrium. With a constant 50% coverage, the high-risk HPV prevalence in women younger than 30 years decreases from 12.9% to 10.4% (Fig. 4A). In women aged 30–60 years, the high-risk HPV prevalence decreases from 5.7% to 4.7%, and in women aged 60+ years, the high-risk HPV prevalence decreases from 1.2% to 0.8%. At 90% coverage, the prevalence of high-risk HPV infection decreases to 9.3% in women younger than 30 years, to 3.9% in women aged 30–60 years, and to 0.6% in women aged 60+ years (Fig. 4B).

The overall prevalence of high-risk HPV infection with (A) 50% and (B) 90% vaccine coverage, and the incidence of cervical cancer with (C) 50% and (D) 90% vaccine coverage, by calendar year in women aged <30 years, 30–60 years, or >60 years. The black curves represent the mean estimates and the gray curves represent the 95% credible intervals. The symbols denote the empirical cervical cancer rate among women aged <30 years (triangles), 30–60 years (open circles), and >60 years (closed circles) over the period 2000–2008. Cervical cancer incidence is underestimated among women aged 60+ years because the model was not calibrated to match birth cohorts with incomplete screening histories.

In our model without vaccination, the number of cervical cancer cases is 548 per 100,000 women per calendar year (95% CI = 421–660). Fewer than 5% of cervical cancer cases occur below the age of 30. Therefore, the vaccination campaign will cause a decline in cervical cancer incidence that becomes noticeable only more than a decade later. Women aged 30–60 years will include those vaccinated from 2024 onward, and by 2054 all screening-eligible women will have been offered vaccination. Hence, the incidence of cervical cancer in women aged 30–60 years declines over this period with half of the benefit achieved by 2035. The decline in cervical cancer incidence is sustained throughout the second half of the 21st century, but additional benefits are restricted to older women (Fig. 4C). With a constant 50% coverage, the total incidence of cervical cancer approaches equilibrium at 290 cases per year (95% CI = 247–343) by the end of the century, a reduction of 47% (95% CI = 41%–48%) relative to the prevaccine equilibrium. At 90% coverage, the equilibrium is 174 (171–176) cases per year. The time scale of the dynamics after vaccination appears insensitive to vaccine coverage (Fig. 4D).

The rate of abnormal cytology in cervical screening is projected to decrease strongly from 2024 onward, when the first cohort of vaccine-eligible women enters the screening-eligible age group. As women are invited to screening once every 5 years, the proportion of abnormal smears decreases in a stepwise manner until 2054, when all women aged 30–60 years will have been offered HPV vaccination. With a constant 50% coverage, the rate of abnormal cytology decreases from 0.040 (95% CI = 0.037–0.042) to 0.032 (0.031–0.033) (Fig. 5A). At 90% coverage, the reduction is to 0.027 (0.027–0.028) (Fig. 5B). The largest reductions in abnormal cytology occur in women aged 30–40 years; hence, the size of the reduction diminishes over time. The age distribution of CIN2/3 has more mass at older age than the age distribution of abnormal cytology, and therefore the decline in the CIN2/3 proportion of abnormal smears is more constant over time, decreasing from 0.34 (0.31–0.36) to 0.25 (0.24–0.26) at a 50% vaccine coverage (Fig. 5C), and to 0.19 at a 90% vaccine coverage (Fig. 5D).

The projected abnormal cytology rate with (A) 50% and (B) 90% vaccine coverage, and the CIN2/3 proportion of abnormal smears with (C) 50% and (D) 90% vaccine coverage, provided that cytologic screening is continued in its current format (7 screening rounds at equidistant intervals from 30 to 60 years) and participation remains constant over time. The black curves represent the mean estimates and the gray curves represent the 95% credible intervals.

Temporal Trends Under Waning Vaccine Efficacy

As a conservative scenario for the long-term impact of HPV vaccination, we also considered a median duration of vaccine protection of only 15 years in combination with a constant low-vaccine uptake (50% of eligible girls). As expected, the reductions in lifetime infection risk for types 16 and 18 are less pronounced. In nonvaccinated women, the lifetime infection risk ultimately drops to a median of 0.57 for HPV-16 (95% CI = 0.39–0.71) and to 0.51 for HPV-18 (0.36–0.62). The hazard of cervical cancer increases relative to the scenario with lifelong vaccine efficacy, especially in older and vaccinated women (eFig. 3A, Moreover, the reduction in high-risk HPV prevalence is apparent only in women younger than 30 years (eFig. 3B,, the decline in cervical cancer incidence is restricted mostly to women aged 30–60 years (eFig. 3C,, and the rates of abnormal cytology and CIN2/3 detection show a rebound after an initial decrease (eFig. 3D,


We have disentangled the direct and indirect effects of HPV vaccination through an assessment of its long-term impact on HPV-16/18 infection rates, cytologic screening outcomes, and cervical cancer incidence as determined by vaccine uptake among preadolescent girls in the Netherlands. For girls vaccinated at age 12, the lifetime risk of cervical cancer drops by 69% if we assume lifelong vaccine efficacy to vaccine-type HPV. This estimated reduction is somewhat lower than the percent of cervical cancers that is attributed to HPV-16 or -18,5 because elimination of vaccine-type HPV unmasks the oncogenic propensity of nonvaccine types. If vaccine uptake in the Netherlands remains at the current level of around 50%, the cervical cancer incidence eventually decreases by 47% with 1 in 4 cancer cases prevented among nonvaccinated women. This reduction is much stronger than the decrease in cervical cancer incidence one would anticipate from naively multiplying 69% by 50%, because our model incorporates the effect of vaccination on transmission of vaccine-type HPV.

To estimate both direct and indirect effects of HPV vaccination, we combined an individual-based simulation model of cervical carcinogenesis with a deterministic HPV transmission model. Unlike previous comparable approaches,27–29 our computations explicitly account for 14 high-risk HPV types. This has several advantages. First, it circumvents the problem of having to extrapolate the impact of HPV vaccination on the total cervical cancer incidence from the impact on cancers attributable to HPV-16 or -18. The multiple high-risk HPV types in the model may invade, persist, and clear independently, with type-specific progression rates up to CIN3 that were informed by observations from a population-based screening trial of HPV DNA testing.25,26 We acknowledge the uncertainty that remains in the model parameterization, but feel that our approach represents an improvement (in terms of estimating the vaccination impact on cancer incidence) over existing dynamic models that focus solely on HPV-16,30 that model a combined HPV-16/18 type,31 or that group all high-risk HPV types other than 16 and 18.27–29,32

Second, our multitype model accounts for multiple type-specific infections without the need for an explicit hierarchical classification of high-risk HPV types by oncogenicity, thereby naturally admitting the possibility that elimination of vaccine-type HPV can unmask underlying high-risk HPV infections or lesions caused by nonvaccine types.16,32 This is a significant improvement, as it gives the ability to estimate the impact of HPV-16/18 vaccination on the occurrence of cervical abnormalities, for which types other than 16 and 18 also bear a large responsibility. This approach allows for a prediction of the impact on screening outcomes, in turn facilitating a better understanding of the economic effects of HPV-16/18 vaccination. Notably, screening outcomes such as abnormal smears and detected CIN2/3 lesions are associated with costs and a temporary decrease in quality of life.

Third, the model allows us to comprehensively account for vaccine cross protection to distinct high-risk HPV types that are genetically related to HPV-16 or -18. Evidence for cross protection is emerging from trials of both the bivalent and the quadrivalent HPV vaccine.7,33 Although we consider it premature to precisely estimate the magnitude of the effect of cross protection on transmission rates and cancer incidence, it would be straightforward to extend vaccine efficacy to high-risk HPV types other than 16 and 18 with our approach.

Our model parameterization is specific to the Netherlands, a country with a well-attended, high-quality cytologic screening program. Our findings are in line with the temporal trends in cancer incidence predicted by other dynamic models of the population impact of HPV vaccination in countries with comparable epidemiologic and screening characteristics, eg, Finland and the United Kingdom.30–32 In countries without organized screening or with a low attendance rate to the screening program, the scope for cervical cancer reduction through HPV vaccination is larger. Furthermore, it is important to consider the geographic variation in high-risk HPV infection risk. In populations with a high prevaccine prevalence of infection, the sexual-contact network might be such that high-risk HPV types spread easily due to high-contact rates between infectious and susceptible persons. Then, the population impact of vaccination will be low compared with that in countries with a low prevalence prior to vaccination.13,14

An important difference between our findings and previously published modeling studies pertains to the possibility of eliminating vaccine-type HPV from the population. We find that elimination of HPV-16/18 is unlikely, while elimination was found to be possible at 70% vaccine coverage in a UK study that considered relatively short duration of natural immunity (up to 3 years).32 In the UK study, scenarios with duration of natural immunity of at least 10 years had a more restricted scope for elimination. The finding that higher coverage is required to eliminate infection in models with longer lasting natural immunity is a well-known phenomenon, as duration of natural immunity is positively correlated with viral transmissibility.34 Partnership-transmission studies have suggested a high transmissibility of HPV-16/18,35 which agrees well with our model parameterization as inferred from the age-specific prevalence of type-specific HPV infection in the prevaccine era.17 Consequently, we estimate that the vaccine types will not be eliminated even at 90% coverage. For men and nonvaccinated women, the risk of HPV-18 infection becomes nonetheless very low if 90% of girls are vaccinated. The finding that HPV vaccination reduces the HPV-18 infection risk more than the HPV-16 infection risk can be explained by the differential prevaccine prevalence of infection, which likely reflects a larger transmission potential of—and hence weaker indirect protection against—type 16 as compared with type 18.

To further reduce the transmission of HPV and to enhance the indirect protection of nonvaccinated women, vaccination of boys might be considered. However, the gain in a further reduction of cervical cancer must be carefully weighed against the extra costs of male vaccination. Moreover, vaccination of girls already lowers the prevalence of HPV 16/18 among boys, although homosexual men will not be protected by a reduced transmission of vaccine types in the heterosexual population. A detailed cost-effectiveness analysis that included numerous HPV-related diseases (related to both men and women) concluded that the inclusion of boys in an HPV vaccination program generally exceeds conventional thresholds of good value for money; expanding vaccine policy to include boys became favorable only in scenarios of low efficacy, low coverage, or low vaccine prices.36

In mathematical models with heterogeneous levels of sexual activity, the incidence of cancer decreases nonlinearly with increasing vaccine coverage.13,14 This nonlinear relation between vaccine coverage and cancer reduction implies diminishing marginal returns from extended uptake in the HPV vaccination program. We predict that the absolute number of cervical cancer cases averted due to herd immunity is highest between 50% and 70% coverage. At these levels of vaccine uptake, indirect effects account for 1 in 4 cancer cases prevented. Our estimate is higher than a previously reported figure of 10% additional reduction in the lifetime risk of HPV-16/18-associated cervical cancer.27,28 This difference is not due to the addition of nonvaccine types to our model; inclusion of nonvaccine types lowers the number of cervical cancers prevented, but the percent of indirectly averted cases among the prevented cancers is not altered. The stronger herd immunity implied in our model could be due to a lower activity or to less heterogeneity in our sexual-contact network, which would be associated with a lower transmission potential and consequently with stronger indirect protection against vaccine-preventable HPV infection.14

We present estimates of cancer incidence that depend explicitly on vaccine coverage and vaccine efficacy, with 95% credible intervals that reflect our uncertainty with regard to viral characteristics of HPV vaccine types. Other sources of uncertainty (in particular, behavioral parameters relating to sexual behavior, screening uptake in vaccinated women, as well as changes in demography) have not entered our calculations. The uncertainty in our model estimates diminished with ongoing calendar time and increasing vaccine coverage—parameters of HPV vaccine types have less bearing on model outcomes when the types in question become obsolete.

Our results suggest that the decline in cervical cancer incidence will be slow, but sustained throughout much of the 21st century if vaccine-induced immunity is not lost over time. We did not observe an association between vaccine coverage and the time needed to reach an impact on cervical cancer incidence; half of the projected benefit in women aged 30–60 years is achieved by 2035, irrespective of vaccine coverage. The impact on cervical cancer precursors will become apparent somewhat earlier, and the full benefit on preinvasive endpoints will be achieved around 2050. The impact of HPV vaccination on abnormal cytology and CIN2/3 detection rates is smaller than the impact on the cervical cancer incidence due to the larger contribution of nonvaccine HPV types to the precancer endpoints.37,38 It follows that the predictive value of abnormal cytology will change due to HPV vaccination, which is important for the future of cervical screening. Wide-scale immunization may also be expected to increase the average age at infection and consequently shift the distribution of cases to older ages. Our results show that, although the peak incidence of infection and of cervical cancer will not shift to older age, the number of cervical cancer cases will decline most strongly between 30 and 40 years of age as a result of HPV vaccination. Altogether, these findings could have major implications for ways in which screening programs may need to be adjusted in vaccinated populations.39–41

Our analysis shows that vaccine-uptake rates achieved so far in the Netherlands yield maximum benefit of herd immunity, but increased vaccine coverage is needed to achieve maximum reduction in cervical cancer incidence. Promotion of the awareness of the link between HPV and cancer for encouraging vaccine uptake should, however, not obscure the importance of secondary prevention efforts. Conveying this double message of adhering to both the immunization program and the cervical screening program is crucial for successful implementation of HPV vaccination.


We thank two anonymous reviewers for constructive comments.


1. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol. 1999;189:12–19.
2. Bosch FX, Lorincz A, Muñoz N, et al. The causal relation between human papillomavirus and cervical cancer. J Clin Pathol. 2002;55:244–265.
3. Muñoz N, Bosch FX, de Sanjosé S, et al. Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med. 2003;348:518–527.
4. Schiffman M, Castle PE, Jeronimo J, et al. Human papillomavirus and cervical cancer. Lancet. 2007;370:890–907.
5. Li N, Franceschi S, Howell-Jones R, Snijders PJ, Clifford GM. Human papillomavirus type distribution in 30,848 invasive cervical cancers worldwide: variation by geographical region, histological type and year of publication. Int J Cancer. 2011;128:927–935.
6. Ault KA; Future II Study Group. Effect of prophylactic human papillomavirus L1 virus-like-particle vaccine on risk of cervical intraepithelial neoplasia grade 2, grade 3, and adenocarcinoma in situ: a combined analysis of four randomised clinical trials. Lancet. 2007;369:1861–1868.
7. Paavonen J, Naud P, Salmerón J, et al. Efficacy of human papillomavirus (HPV)-16/18 AS04-adjuvanted vaccine against cervical infection and precancer caused by oncogenic HPV types (PATRICIA): final analysis of a double-blind, randomised study in young women. Lancet. 2009;374:301–314.
8. National Institute of Public Health and the Environment (RIVM). Information on the cervical cancer vaccine [RIVM web site]. February, 2009. Available at: Accessed October 28, 2010.
9. Boot HJ, Wallenburg I, de Melker HE, et al. Assessing the introduction of universal human papillomavirus vaccination for preadolescent girls in The Netherlands. Vaccine. 2007;25:6245–6256.
10. Coupé VM, van Ginkel J, de Melker HE, Snijders PJ, Meijer CJ, Berkhof J. HPV16/18 vaccination to prevent cervical cancer in The Netherlands: model-based cost-effectiveness. Int J Cancer. 2009;124:970–978.
11. De Kok IM, van Ballegooijen M, Habbema JD. Cost-effectiveness analysis of human papillomavirus vaccination in the Netherlands. J Natl Cancer Inst. 2009;101:1083–1092.
12. Rogoza RM, Westra TA, Ferko N, et al. Cost-effectiveness of prophylactic vaccination against human papillomavirus 16/18 for the prevention of cervical cancer: adaptation of an existing cohort model to the situation in the Netherlands. Vaccine. 2009;27:4776–4783.
13. Hughes JP, Garnett GP, Koutsky L. The theoretical population-level impact of a prophylactic human papilloma virus vaccine. Epidemiology. 2002;13:631–639.
14. Garnett GP, Kim JJ, French K, Goldie SJ. Modelling the impact of HPV vaccines on cervical cancer and screening programmes. Vaccine. 2006;24(suppl 3):S3/178–S3/186.
15. Kim JJ, Brisson M, Edmunds WJ, Goldie SJ. Modelling cervical cancer prevention in developed countries. Vaccine. 2008;26(suppl 10):K76–K86.
16. Choi YH, Jit M, Gay N. Unmasking HPV genotypes as impact of HPV vaccination. In: Abstract Book 25th International Papillomavirus Conference; 2009; Malmo, Sweden. P-29.44.
17. Bogaards JA, Xiridou M, Coupé VM, Meijer CJ, Wallinga J, Berkhof J. Model-based estimation of viral transmissibility and infection-induced resistance from the age-dependent prevalence of infection for 14 high-risk human papillomavirus types. Am J Epidemiol. 2010;171:817–825.
18. Liaw KL, Hildesheim A, Burk RD, et al. A prospective study of human papillomavirus (HPV) type 16 DNA detection by polymerase chain reaction and its association with acquisition and persistence of other HPV types. J Infect Dis. 2001;183:8–15.
19. Rousseau MC, Pereira JS, Prado JC, Villa LL, Rohan TE, Franco EL. Cervical co-infection with human papillomavirus (HPV) types as a predictor of acquisition and persistence of HPV infection. J Infect Dis. 2001;184:1508–1517.
20. Bulk S, van Kemenade FJ, Rozendaal L, Meijer CJ. The Dutch CISOE-A framework for cytology reporting increases efficacy of screening upon standardisation since 1996. J Clin Pathol. 2004;57:388–393.
21. De Carvalho N, Teixeira J, Roteli-Martins CM, et al. Sustained efficacy and immunogenicity of the HPV-16/18 AS04-adjuvanted vaccine up to 7.3 years in young adult women. Vaccine. 2010;28:6247–6255.
22. De Graaf H, Meijer S, Poelman J, Vanwesenbeeck I. Sex Below the Age of 25: Sexual Health of Youth in the Netherlands in 2005 [in Dutch]. Delft, The Netherlands: Eburon Academic Publishers; 2006. Rutgers Nisso Groep studies, no. 7.
23. Bakker F, Vanwesenbeeck I, eds. Sexual Health in the Netherlands in 2006 [in Dutch]. Delft, The Netherlands: Eburon Academic Publishers; 2006. Rutgers Nisso Groep studies, no. 9.
24. Berkhof J, Bulkmans NW, Bleeker MC, et al. Human papillomavirus type-specific 18-month risk of high-grade cervical intraepithelial neoplasia in women with a normal or borderline/mildly diskaryotic smear. Cancer Epidemiol Biomarkers Prev. 2006;15:1268–1273.
25. Bulkmans NW, Berkhof J, Rozendaal L, et al. Human papillomavirus DNA testing for the detection of cervical intraepithelial neoplasia grade 3 and cancer: 5-year follow-up of a randomised controlled implementation trial. Lancet. 2007;370:1764–1772.
26. Berkhof J, De Bruijne MC, Zielinski GD, Meijer CJ. Natural history and screening model for high-risk human papillomavirus infection, neoplasia and cervical cancer in the Netherlands. Int J Cancer. 2005;115:268–275.
27. Kim JJ, Andres-Beck B, Goldie SJ. The value of including boys in an HPV vaccination programme: a cost-effectiveness analysis in a low-resource setting. Br J Cancer. 2007;97:1322–1328.
28. Goldie SJ, Kim JJ, Kobus K, et al. Cost-effectiveness of HPV 16, 18 vaccination in Brazil. Vaccine. 2007;25:6257–6270.
29. Kim JJ, Goldie SJ. Health and economic implications of HPV vaccination in the United States. N Eng J Med. 2008;359:821–832.
30. Barnabas RV, Laukkanen P, Koskela P, Kontula O, Lehtinen M, Garnett GP. Epidemiology of HPV 16 and cervical cancer in Finland and the potential impact of vaccination: mathematical modelling analyses. PLoS Med. 2006;3:e138.
31. Elbasha EH, Dasbach EJ, Insinga RP. Model for assessing human papillomavirus vaccination strategies. Emerg Infect Dis. 2007;13:28–41.
32. Choi YH, Jit M, Gay N, Cox A, Garnett GP, Edmunds WJ. Transmission dynamic modelling of the impact of human papillomavirus vaccination in the United Kingdom. Vaccine. 2010;28:4091–4102.
33. Brown DR, Kjaer SK, Sigurdsson K, et al. The impact of quadrivalent human papillomavirus (HPV; types 6, 11, 16, and 18) L1 virus-like particle vaccine on infection and disease due to oncogenic nonvaccine HPV types in generally HPV-naive women aged 16–26 years. J Infect Dis. 2009;199:926–935.
34. Garnett GP. Role of herd immunity in determining the effect of vaccines against sexually transmitted disease. J Infect Dis. 2005;191(suppl 1):S97–S106.
35. Burchell AN, Tellier PP, Hanley J, Coutlée F, Franco EL. Human papillomavirus infections among couples in new sexual relationships. Epidemiology. 2010;21:31–37.
36. Kim JJ, Goldie SJ. Cost-effectiveness analysis of including boys in a human papillomavirus vaccination programme in the United States. BMJ. 2009;339:b3884.
37. Bosch FX, Burchell AN, Schiffman M, et al. Epidemiology and natural history of human papillomavirus infections and type-specific implications in cervical neoplasia. Vaccine. 2008;26(suppl 10):K1–K16.
38. Cuzick J, Castañón A, Sasieni P. Predicted impact of vaccination against human papillomavirus 16/18 on cancer incidence and cervical abnormalities in women aged 20–29 in the UK. Br J Cancer. 2010;102:933–939.
39. Myers E, Huh WK, Wright JD, Smith JS. The current and future role of screening in the era of HPV vaccination. Gynecol Oncol. 2008;109(suppl 2):S31–S39.
40. Franco EL, Mahmud SM, Tota J, Ferenczy A, Coutlée F. The expected impact of HPV vaccination on the accuracy of cervical cancer screening: the need for a paradigm change. Arch Med Res. 2009;40:478–485.
41. Coupé VM, de Melker HE, Snijders PJ, Meijer CJ, Berkhof J. How to screen for cervical cancer after HPV16/18 vaccination in The Netherlands. Vaccine. 2009;27:5111–5119.

Supplemental Digital Content

© 2011 Lippincott Williams & Wilkins, Inc.