SEARCH

SEARCH BY CITATION

Keywords:

  • simulation model;
  • human papillomavirus;
  • cervical cancer;
  • vaccination

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We evaluated the cost-effectiveness of HPV16/18 vaccination for girls aged 12 years in The Netherlands in addition to cervical cancer screening. For this purpose, we developed a simulation model that describes the relation between each of the high-risk human papillomavirus (hrHPV) types and cervical disease, allowing the occurrence of multiple type-specific infections. Model parameters were derived from Dutch cohort studies, including a large population-based screening trial, and from the national cervical cancer registry. The model satisfactorily reproduced Dutch data on HPV infection and the presence of cervical lesions. For our base-case scenario in which 85% of the girls aged 12 years were vaccinated against types 16/18 (95% efficacy, lifelong protection), the model predicted a decrease of 60% in the number of cervical cancer cases and cervical cancer deaths indicating that substantial health benefits can be achieved. Health savings were robust against changes in the vaccine efficacy (varied from 85% to 98%) but savings showed a substantial reduction when the efficacy started waning 10 years after vaccination. The discounted costs per quality-adjusted life year (QALY) were € 19,500/QALY (range € 11,000 to € 25,000/QALY) and lied near the cost-effectiveness threshold of € 20,000/QALY used in The Netherlands. The simulations further showed that vaccination cannot replace screening because vaccination without screening was less effective than screening in preventing cancer in women over 40 years of age. In conclusion, our model results support the implementation of HPV16/18 vaccination in young women in addition to cervical cancer screening. © 2008 Wiley-Liss, Inc.

Infection with high-risk human papillomavirus (hrHPV) is the necessary cause of cervical cancer.1 Recently, prophylactic vaccines have come available that protect against infection with the oncogenic HPV types 16 and 18.2, 3 HPV 16 and/or 18 (HPV 16/18) is present in 60–80% of all cervical cancer cases.4, 5 Therefore, mass vaccination may have a substantial impact on the burden of cervical cancer, even in countries with organized cervical screening. The safety and efficacy of HPV16/18 vaccination has been demonstrated in several large randomized trials.3, 6, 7 For women without detectable HPV16/18 DNA, an efficacy of 90–98% against HPV16/18 positive cervical intraepithelial neoplasia grade 2 and 3 (CIN2/3) was reported. In addition, some evidence for cross-protection against other HPV types was reported.3 Because of the limited follow-up, long-term efficacy is still uncertain but within the first 5 years no reduction in efficacy was observed and HPV type-specific antibody levels remained at a high level.8

To decide whether or not to introduce nationwide HPV16/18 vaccination, insight into the cost-effectiveness of vaccination is needed in addition to information on safety and efficacy. For this purpose, simulation models have been developed that predict country-specific costs and health benefits. Model-based cost-effectiveness studies have been carried out for the US,9–12 the UK,13 and Canada.14 Although the results of these studies vary widely, the majority of the studies suggest that HPV16/18 vaccination can be implemented in a cost-effective manner. To predict the cost-effectiveness of HPV vaccination in The Netherlands where the incidence of cervical cancer is already very low, we developed a simulation model which describes the relation between 14 high-risk human papillomavirus (hrHPV) types and cervical disease. The model allows for the occurrence of multiple type-specific infections, which are frequently observed even in women older than 30 years.15 The incidences of type-specific HPV infections and CIN2/3 were derived from Dutch cohort studies including a population-based screening trial of 44,102 women.16

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Model

We developed a simulation model for the natural history of a high-risk HPV infection and the progression into cervical abnormalities and cervical cancer. The model simulates health trajectories of a cohort of 12-year old girls until they are deceased. The preinvasive part of the model consists of 14 parallel Markov chains corresponding to an infection with HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68. In the model, a woman may have multiple infections at a certain point in time and for example have a HPV18-positive CIN2 and a HPV16-positive CIN3. We categorized women as high risk, intermediate risk, or low risk for acquiring a HPV infection (85.1% low risk, 13.7% intermediate risk, 1.2% high risk) with HPV incidences in the high-risk group that were 30-fold higher than in the low-risk group. The categorization was based on a cluster analysis of HPV typing data.17 Notably, a model without different risk groups would underestimate the number of multiple infections.

The structure of the natural history model is shown in Figure 1. Women without HPV infection may develop CIN1. Until state CIN3, spontaneous clearance of a HPV infection and regression of cervical lesions is possible. After HPV clearance, women are again susceptible for infection. The state “cervical cancer” consists of states FIGO1 and FIGO2+. In addition to the states shown in Figure 1, the model contains states for survival of cervical cancer, death through cervical cancer, death through other causes and hysterectomy. The data used for estimating the probability of each state transition are shown in Table I.

thumbnail image

Figure 1. Simplified flowchart of the model.

Download figure to PowerPoint

Table I. Model Parameters, Screening and Vaccine Characteristics, Quality of Life, and Costs
State transition Six monthly transition probability1References
  • 1

    The reported ranges reflect the probability variation with age.

No HPV Infection to CIN0, HPV+Type 160.000–0.04017
Type 180.000–0.009
Type 310.000–0.016
Type 330.000–0.005
Type 450.000–0.007
Other types0.000–0.009
CIN0, HPV+ to No HPV infectionAll types0.3915
CIN0, HPV+ to CIN1, HPV+All types 18
 <30 years0.1–0.10
 ≥30 years0.10
CIN1, HPV+ to CIN2, HPV+Type 160.4019,20
Type 18/330.25
Type 310.15
Other types0.10
CIN2, HPV+ to CIN3, HPV+Type 160.4019,20
Type 18/330.25
Type 310.15
Other types0.10
CIN0, HPV+ to CIN3, HPV+Type 16 21,22
 <30 years0.002–0.02
 ≥30 years0.02
Type 18/31/33 
 <30 years0.0005–0.005
 ≥30 years0.005
Other types0
CIN1, HPV+ to No HPV InfectionAll types0.2523–25
CIN2, HPV+ to No HPV InfectionAll types0.2523–25
CIN3, HPV+ to No HPV InfectionAll types0.3023–25
No HPV Infection to CIN1, HPV−All types0.00316
CIN1, HPV− to No HPV InfectionAll types0.6024
CIN3, HPV+ to Pre1FIGO1Type 160.2026,27
Type 18/450.40
Other types0.10
FIGO Stage 1 to FIGO Stage 2+ 0.04828
FIGO Stage 1 to detected FIGO1 0.032 
FIGO Stage 2+ to detected FIGO2+ 0.30 
Screening characteristics Parameter valuesReferences
Screening age (years) 30–60 years 
Length of screening interval 5 years 
Proportion of women who never attend screening 10%29
Compliance per screening round 80%16
Positive smear rate in
 CIN0 0.01516
 CIN1 0.40 
 CIN2 0.50 
 CIN3 0.75 
Vaccine characteristics Parameter valuesReferences
Vaccine efficacy 95% 
Age at vaccination 12 years 
Vaccine coverage 85% 
Health stateDuration (years)Quality of life estimate 
Abnormal smear0.50.99510,30
CIN1 treatment0.50.97 
CIN2/3 treatment0.50.93 
CIN2/3 residual0.50.93 
Detection of FIGO 10.50.65 
Detection of FIGO 2+0.50.55 
Follow-up FIGO14.50.97 
Follow-up FIGO2+4.50.85 
Procedures Costs (€) (2006)References
Conventional cytological screening  31
 First smear 42.10 
 Repeat smear 45.40 
 Administrative costs 5.70 
Vaccination
 Vaccine 3 doses (€125 per dose and €6 administration costs) 393.00 
 Booster dose (€125 per dose and €6 delivery costs) 131.00 
 Administrative costs (full vaccination course) 22.50 
Diagnosis, treatment, follow-up  31,32
 CIN0 305.61 
 CIN1 1325.35 
 CIN2 1524.98 
 CIN3 1656.69 
 FIGO Stage 1 9046.97 
 FIGO Stage 2+ 10659.41 
 Palliative care≤50 years42,428 
50–70 years30,242 
≥70 years12,873 

The HPV type incidences and the development of preinvasive lesions were estimated from data collected in a large screening trial16, 17 and clinical cohorts.24, 25, 33, 34 Progression probabilities from CIN1 to CIN3 and from CIN3 to cancer were type-specific.5, 19, 27 The progression rate from HPV infection to CIN3 was age-dependent and assumed to be 2.5-fold higher above the age 30 than at young age.35 The progression from CIN3 to cervical cancer was calibrated on registry data (www.ikcnet.nl) and on historical data.36 The mean duration from CIN3 to microinvasive cervical cancer was set at 14 years.37 To accurately model this long duration, 2 intermediate tunnel states were included in the model (Pre1FIGO1 and Pre2FIGO1 in Table I).38 The progression between cervical cancer states was estimated from hospital data on clinical cancer cases.28 Stage-specific survival of cervical cancer was based on registry data (www.ikcnet.nl). Dutch hysterectomy rates and mortality rates were obtained from van Ballegooijen et al. (1993).39

Screening

The screening module implements the current Dutch guidelines on cervical screening. That is, women between 30 and 60 years of age are invited to cytological screening once every 5 years. The referral policy is to send a woman to the gynecologist for colposcopy and a biopsy if the smear test result is moderate dyskaryosis or worse or if a borderline/mild dyskaryotic smear is followed by a second abnormal smear at 6 or 18 months. Cervical smears are read according to the CISOE-A classification, and interpreted as normal (Pap1), borderline or mild dyskaryosis (Pap2/3a1), moderate dyskaryosis or worse (Pap3a2 and higher). A translation of the CISOE-A classification into the Bethesda 2001 classification is available.40 The screening parameter values are shown in Table I.

Vaccination

We assumed that 85% of all 12-year-old girls would receive the 3 vaccination doses based on communication with immunization experts in the Netherlands. We assumed a vaccine efficacy of 95% against incident infection with types 16 and 18.3, 6, 7, 41, 42 Because antibody levels in the studies remained high during the 5-year follow-up period, we assumed lifelong protection of the vaccine. We separately investigated whether administering a booster at 30 years of age influenced the conclusions of our cost-effectiveness analysis.

Costs and quality-adjusted life-years

The costs in euros per unit of health care resource utilization are presented in Table I. All costs were indexed at year 2006. The costs of screening and treatment were published previously and were updated to 2006 using the consumer price index.31, 32, 43 The utilities for different health states (Table I) were based on international publications.10, 30 Following the Dutch guidelines, the discounting rate per year for costs and health effects were set at 4% and 1.5%, respectively.

Analyses

Model predictions were obtained by simulating the health trajectories of a cohort of 10,000,000 women from age 12 until age 100. Results were divided by 100 to obtain predictions for a cohort of 100,000 women. This roughly corresponds to the age-cohort size of girls in The Netherlands.

We simulated the following scenarios: (I) no screening or vaccination, (II) screening but no vaccination, (III) vaccination but no screening, (IV) screening and vaccination. For each scenario, we determined the number of detected CIN2/3 lesions, the number of cervical cancer cases, the number of cervical cancer deaths, the total costs and quality-adjusted life-years (QALYs). We computed the incremental cost effectiveness ratio (ICER) of adding vaccination to screening by dividing the difference in costs between scenarios IV and II by the difference in QALYs.

Sensitivity analyses

We studied the robustness of the cost-effectiveness of adding vaccination to screening in comparison with screening only with regard to changes in the vaccine efficacy [98% (Scenario V), 90% (Scenario VI) and 85% (Scenario VII)]. Furthermore, we studied the effects of waning efficacy: 95% efficacy for 10 years, exponential decrease in efficacy of 50% during each following 20 years (Scenario VIII: slow waning), 10 years (Scenario IX: intermediate waning), or 5 years (Scenario X: fast waning). The effect of cross-protection of the vaccine against HPV types 31 and 45 was evaluated by assuming an efficacy against both types of either 90% (Scenario XI: high cross-protection) or 50% (Scenario XII: low cross-protection). Finally, we studied the effect of differences in screening compliance in vaccinated versus nonvaccinated women: 70% versus 80% attendance per round in vaccinated and nonvaccinated women, respectively (Scenario XIII) and 20% versus 10% nonattendance in combination with 70% versus 80% attendance per round in vaccinated versus nonvaccinated women, respectively (Scenario XIV).

Finally, we studied the sensitivity of the cost-effectiveness results with regard to changes in the disease parameters. We only considered the disease parameters HPV incidence at young age and the durations from state CIN0 to CIN3, CIN3 to FIGO 1 and FIGO 1 to FIGO 2+. For those disease parameters, information is very limited. We increased the type-specific HPV incidence at young age by setting the peak incidence at age 15 years. Each of the durations was increased and decreased by factor 1.5. To retain a model that does not contradict available data, a change in 1 disease parameter should be compensated for by a change in another disease parameter. We compensated the increase in the type-specific HPV incidence by increasing the HPV clearance rate and we compensated the change in the durations by changing the corresponding progression probabilities. The level of compensation was determined by minimizing the deviance from the calibration targets HPV prevalence, CIN2/3 prevalence and cancer incidence.

The deviances from the calibration targets were measured by the sum of squared differences between age-specific model predictions on the one hand and the HPV and CIN2/3 prevalence data from the POBASCAM study and the cancer incidence from the national cancer registry on the other hand. A model with varied disease parameters was considered acceptable if the ratio of the deviance of that model and the deviance of the base-case model was not larger than 1.25 for all 3 calibration targets. We computed ICERs of vaccination in addition to screening for all acceptable models.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Model predictions for the current situation with screening

In Figure 2 and in Table II, model predictions for the current situation of 5-yearly screening of women between age 30 and 60 are compared to the POBASCAM study and cancer incidences from the national registry, Figure 2a shows the type-specific prevalences for types 16, 18, 31, 33 and 45. Above 30 years of age, predicted and observed prevalences agree well. For younger ages, the predictions may deviate somewhat from the observed prevalences, because for young women limited HPV type data were available.

thumbnail image

Figure 2. Type-specific HPV prevalence (a), the proportion of screened women with a detected CIN2/3 lesions (b) and the incidence of cervical cancer per 100,000 women years (c). The black lines show the predictions by the model. The grey lines or marks represent either the POBASCAM data (a and b) or the data from the Dutch cancer registry between 1995 and 2003 (c).

Download figure to PowerPoint

Table II. The HPV Type Distribution in Cervical Cancer: Model Predictions and International Literature
HPV typeModel with screeningRef. 26Ref. 4Ref. 5
HPV1662.8%69.9%65.1%54.3%
HPV1810.8%12.0%8.1%12.6%
HPV319.3%7.2%5.8%4.2%
HPV335.7%3.6%1.2%4.3%
HPV454.6%6.0%2.3%4.2%

The predicted proportion of women with CIN2/3 agrees well with the proportion observed in the POBASCAM study (Figure 2b). The age-specific cervical cancer incidence as predicted by the model follows the data from the national cancer registry (www.ika.nl) rather well (Figure 2c). Note that the observed patterns from the national registry tend to be flatter than the predicted cervical cancer incidence.

Finally, Table II shows the HPV type distribution in cervical cancer as predicted by the model in comparison with both Dutch HPV typing data26 and international typing data.4, 5 The model predictions are in agreement with the published data.

Effects of screening and vaccination

Table III gives the model predictions of clinical outcome measures for (I) no screening or vaccination, (II) screening only, (III) vaccination only and (IV) both vaccination and screening. Predictions are given for a cohort of 100,000 women. The model predicts that screening is more effective than vaccination in preventing cancer (634 vs. 731 cervical cancer cases). Moreover, the cervical cancer death rate is higher under vaccination (277/731 = 38%) than under screening (184/634 = 29%) The model also predicts that vaccination reduces the number of cervical cancer cases and mortality cases in a population that receives screening by 61%.

Table III. The Effect of Screening, Vaccination, and the Combination of Screening and Vaccination in a Cohort of 100,000 Women Simulated from 12 to 100 Years on the Number of Cervical Cancer Cases and the Number of Cervical Cancer Deaths
ScenarioLife-time number of cervical cancer casesLife-time number of deaths due to cervical cancer
Scenario I: No vaccine, no screening1851699
Scenario II: Screening only634184
Scenario III: Base case vaccination only (95% efficacy type 16/18, 85% coverage, lifelong protection)731277
Scenario IV: Base case vaccination and screening24771

Figure 3 shows the predicted age-specific cervical cancer incidence for the 4 scenarios. It can be seen that vaccination prevents more cervical cancer at young age (younger than 40) whereas screening has a strong effect on the cervical cancer cases above 40 years of age. Furthermore, Figure 3 shows that, compared to screening alone, the addition of vaccination leads to a larger reduction in cervical cancer at young age (64% between 35 and 39) than at older age (55% between 80 and 84).

thumbnail image

Figure 3. The predicted incidence of cervical cancer in a scenario without screening or vaccination (I), a scenario with screening only (II), a scenario with vaccination only (III) and a scenario with vaccination and screening (IV).

Download figure to PowerPoint

The cost-effectiveness of vaccination in addition to screening

Table IV and Figure 5 show the effects and costs of screening scenario II and 9 scenarios combining vaccination and screening (IV–XIV). For our base-case vaccination + screening scenario IV, a decrease of 61% in the incidence of cervical cancer (247/634) and in cervical cancer deaths (71/184) was found in comparison with screening scenario II. Furthermore, we found an increase in discounted quality adjusted life years (QALYs) of 0.0167 and an increase in discounted costs of € 324.47 for scenario IV compared to II which means that the incremental cost-effectiveness ratio (ICER) equals € 19,429 per QALY gained. This value lies near the cost-effectiveness threshold of € 20,000/QALY for preventive interventions in The Netherlands. The effect of the vaccine price on the ICER of our base-case vaccination scenario is illustrated in Figure 4. If the vaccine price is lowered to € 75 per dose, then the ICER decreases to a cost-effective value of € 10,000/QALY. If a booster at age 30 is required to obtain lifelong protection, the ICER is € 23,000/QALY for a cost price of € 125 per dose. The ICER for the booster scenario will become lower than the Dutch threshold value of € 20,000/QALY when the dose price lies below € 100.

thumbnail image

Figure 4. The effect of the vaccine price per dose on the cost-effectiveness of vaccination in case no booster is given (dots) or in case one booster is given at 30 years of age (boxes).

Download figure to PowerPoint

thumbnail image

Figure 5. The effect of different assumptions for the vaccine characteristics on the incidence of cervical cancer. Predictions for screening Scenario II and base-case vaccination Scenario IV are shown for comparison.

Download figure to PowerPoint

Table IV. Cost-Effectiveness, Costs, and Effects of Various Scenarios of HPV Vaccination in a Cohort of 100,000 Women Simulated from 12 to 100 Years on the Number of Cervical Cancer Cases, the Number of Cervical Cancer Deaths
ScenarioLife-time number of cervical cancer casesLife-time number of deaths due to cervical cancerUndiscountedDiscounted (4% for costs and 1.5% for effects)
QALYCosts (€)QALYCosts (€)ICER
Scenario II: Screening only63418468.7050456.3442.6908139.73Reference
Scenario IV: Base case vaccination (95% efficacy type 16/18, 85% coverage, lifelong protection) and screening2477168.7405706.9142.7075464.1919.429
Scenario V: High-efficacy vaccination (98% efficacy)2306668.7421702.4442.7082462.9418.554
Scenario VI: Intermediate-efficacy vaccination (90% efficacy)2757968.7383713.8942.7065466.1920.794
Scenario VII: Low-efficacy vaccination (85% efficacy)2968568.7363719.9542.7055468.0922.337
Scenario VIII: Slow waning vaccination (after 10 years exponential decrease of 50% each 20 years)35910568.7335734.8742.7045470.1724.120
Scenario IX: Intermediate waning vaccination (after 10 years exponential decrease of 50% each 10 years)40812068.7298749.2442.7028474.0627.862
Scenario X: Fast waning vaccination (after 10 years exponential decrease of 50% each 5 years)46113768.7243765.7242.7004478.8535.325
Scenario XI: Low cross-protection vaccination (50% efficacy type 31/45)2236468.7427698.0642.7085461.6118.186
Scenario XII: High cross-protection vaccination (90% efficacy type 31/45)2015968.7440690.8342.7091459.3817.467
Scenario XIII: Low attendance per screening round after vaccination (70% attendance per round)2627768.7390679.1042.7068452.8119.568
Scenario XIV: High nonattendance and low attendance per screening round after vaccination (20% nonattendance and 70% attendance per round)2969168.7348665.4042.7051447.5421.525

Sensitivity analyses

The effects of varying the assumptions about the vaccine efficacy and the screening compliance after vaccination on the cost-effectiveness results are presented in Table IV and Figure 5. The vaccine efficacy level has only a limited effect on the results. With an efficacy of 85%, the model still predicts a 53% reduction in cervical cancer incidence and deaths when compared to screening only (II). Waning efficacy, on the other hand, has a strong effect on the results. Even for scenario VIII with “slow waning,” the reduction in cervical cancer cases compared to screening only is as low as 43% whereas a reduction of 61% is predicted for scenario IV with lifelong protection. For the slow, moderate and fast waning scenarios (VIII, IX and X), the ICER increases to € 24,000, € 28,000 and € 35,000/QALY, respectively. Assuming cross-protection against types 31 and 45 (XI and XII) improves the clinical outcomes (reductions in cervical cancer cases of 65% and 68%, respectively) and cost-effectiveness (€ 18,000 and € 17,500/QALY, respectively) in comparison with the base-case vaccination scenario (IV), but the effect is small. Assuming a 10% decrease in attendance per screening round for vaccinated women compared to nonvaccinated women (scenario XIII) hardly affects the cost-effectiveness results. Increasing the proportion of vaccinated women that never attends screening from 10 to 20% (scenario XIV) leads to a small decrease in the effect of vaccination (number of deaths increases from 71 to 91) and a small increase in the ICER.

The sensitivity of the cost-effectiveness results with regard to changes in the disease parameters is shown in Figure 6. In total, 39 models yielded satisfactory predictions of the calibration targets HPV prevalence, CIN2/3 prevalence and incidence of cervical cancer. Figure 6 depicts a cost-effectiveness plane with the differences in costs and QALYs between our base-case vaccination + screening scenario IV and screening scenario II for all 39 models. It can be seen that most models lead to lower predicted costs and larger predicted effects of adding HPV vaccination to screening compared to our base-case model. The corresponding ICERs vary between € 11,000/QALY and € 25,000/QALY with an average of € 17,000/QALY. Notwithstanding the considerable variation in ICERs, none of the models gave rise to an ICER that was substantially higher than the Dutch cost-effectiveness threshold of € 20,000/QALY.

thumbnail image

Figure 6. Difference in discounted costs and QALYs between the base-case scenario of vaccination and screening compared to the scenario of screening only. Results are shown for the base-case model (black) and for 39 models with sets of varied disease parameters (grey).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We assessed the cost-effectiveness of HPV16/18 vaccination in The Netherlands. Our simulations showed that implementing HPV vaccination alongside cervical screening can lead to a considerable reduction in the incidence and mortality of cervical cancer (approximately 60%). The simulations also showed that vaccination cannot replace screening as vaccination (without screening) led to a strong reduction in cancer at young age but was less effective than screening in preventing cancer in women over 40 years of age. Besides, screening averted more cancer deaths than vaccination. This supports the idea of using vaccination to prevent cancer in young women and screening to prevent cancer in older women.44 It can further be noted that the attractiveness of HPV vaccination not only lies in its ability to prevent rather than cure cervical disease but also in its expected accessibility. At present, around 10% of Dutch women never attend cervical screening.29 This group of women has a relatively high risk to develop cervical cancer. Possibly, these women are easier to reach through vaccination than through screening.

Cost calculations indicated that at the current price indication of € 125 per dose, HPV16/18 vaccination of 85% percent of 12-year-old girls (95% efficacy, lifelong protection) is likely to be a cost-effective addition to screening as the computed ICER of € 19,500 per QALY lied close to the current cost-effectiveness threshold in The Netherlands of € 20,000 per QALY. If a booster at age 30 years will be required to achieve lifelong protection, the cost price per dose should decrease to € 100 per dose to reach the ICER threshold of € 20,000 per QALY.

An important strength of our simulation model is that it is based on data from large preclinical and clinical cohorts in The Netherlands, including data from a screening cohort of 45,102 Dutch women.16 As such, it was possible to estimate the age-dependent prevalences of hrHPV types and the progression rates of type-specific hrHPV infections and preclinical stages of cervical cancer from Dutch data. Furthermore, our model differs from other models in that it consists of parallel Markov chains, one for each HPV type. This makes it possible to model the occurrence of multiple infections, which is a common phenomenon. We explored the influence of the occurrence of multiple infections on the cost-effectiveness results by repeating model simulations for a population in which every women has the same age-dependent risk of acquiring a HPV infection (one risk group rather than three). The model with one risk group has similar HPV-type specific prevalences as the three group model but has a lower prevalence of multiple infections (less than 5% for all ages). In the one risk group model, the ICER of vaccination in addition to screening is lower than in the 3 group model (€ 18,000 per QALY versus € 19,500 per QALY). Therefore, if the number of multiple infections is underestimated by the model, cost-effectiveness calculations for vaccination in addition to screening may be too optimistic. We did not explore the possible influence of coinfection with HPV16/18 on the progression risk of non-HPV16/18 infections. We decided this because a clear association between the clearance and persistence of infection by non-HPV16/18 types and the presence of either HPV16 or 18 has not been found.15, 45, 46 Nonetheless, if non-HPV16/18 types were less likely to progress to cancer in the absence of HPV16/18, the effects of vaccination would be stronger than in our model and the ICER of vaccination in addition to screening would drop.

For our calculations, we made the following somewhat conservative assumptions. First, we assumed that vaccination coverage is 85%. For young children, coverage rates of over 90% are reported in the national immunization program. We assumed a lower coverage rate because the age of vaccination is higher than for other vaccines in the immunization program and because 3 doses of the vaccine need to be administered. Second, the effect of herd immunity is not incorporated in our simulation model. Herd immunity refers to the phenomenon that unvaccinated persons are (partly) protected against infection because vaccinated persons cannot contribute to spreading the virus. At the current assumed compliance, herd immunity leads to a higher cost-effectiveness of vaccination. Third, in our base-case vaccination scenario, no cross-protection against other HPV types is assumed. Cross-protection against HPV types 31 and 45 has been reported,3 but the effect on the incidence of cervical lesions is unclear. We explored the effect of 50% and 90% cross-protection against HPV31 and 45, respectively, and found that it led to a slight improvement in the cost-effectiveness of vaccination (decrease in ICER of € 1000–€ 2000/QALY). Fourth, we did not include the potential effect of HPV vaccination on other HPV-related cancers, such as vulva, anus and oropharynx carcinoma. Together, the incidence of these types of cancer that is attributable to HPV16 and 18 is around 40% of the incidence of cervical cancer.47 Although hardly any evidence is available on this topic, it may be expected that HPV vaccination also offers protection against these types of cancer. Finally, we did not consider the benefit of protecting against HPV6 and HPV11 infections, even though 1 of the 2 available vaccines offers protection against these 2 low-risk types. HPV 6 and 11 cause genital warts and also occasionally cause the rare but severe condition of recurrent respiratory papillomatosis. Although genital warts are not malign, treatment costs in The Netherlands are about 1 million euro per year.48, 49 Therefore, offering protection against HPV6 and HPV11 in addition to HPV16 and HPV18 may lead to a further improvement of the cost-effectiveness of the vaccine. However, it must also be realized that the success of implementing mass HPV vaccination will eventually depend on whether the vaccine can offer long-term protection against cervical cancer.

Together with a number of conservative assumptions, we optimistically assumed that HPV16/18 vaccination offers lifelong protection. Currently, the follow-up of long-term efficacy studies provide evidence for a high efficacy for up to 5 years.3, 6, 7 In addition, it has been shown that HPV vaccination leads to high type-specific antibody levels, with a plateau between 24 and 60 months after vaccination and that immune memory is induced as well.8, 50 Although extrapolation of HPV antibody levels beyond clinical data suggests long-term protection,51 life-long protection cannot be guaranteed. Therefore, we investigated in our simulation study the effects of waning vaccine efficacy over time. We showed that waning of vaccine efficacy leads to a lower reduction in lesions and cancer cases, resulting in less favorable cost-effectiveness results (between € 24,000/QALY and € 35,000/QALY). To obtain an acceptable protection against disease a booster may be necessary and therefore we computed the cost-effectiveness of vaccination if one booster is administered at 30 years of age (€ 23,000/QALY). It should be noted though that our model overestimates the impact of waning efficacy due to the absence of herd immunity.

It is not easy to compare our results with other model-based studies of the cost-effectiveness of vaccination. There are many reasons, for this among which the different modeling approaches used (e.g., dynamic versus static modeling), the different estimates for the parameters describing the natural history of cervical cancer, the large variety in assumptions for the vaccine characteristics and the variety in reference screening scenarios. Furthermore, in The Netherlands, costs are discounted at 4% and health effects at 1.5% whereas in most countries both cost and effects are discounted at 3%. In particular cost-effectiveness ratios are very sensitive to differences in assumptions. It seems therefore most informative to compare results on costs and health effects separately. Goldie et al.10 presented results for a large number of vaccination scenarios and, even more interesting, compared their results with different reference screening scenarios in which they varied the starting age of screening and the length of the screening interval. Compared to 5-yearly screening from 30 years onward, they calculated that adding HPV vaccination reduced the lifetime risk of cervical cancer with 61% at an incremental discounted lifetime cost of $305. Their reduction in lifetime risk is similar to ours. Kohli et al. estimated that HPV16/18 vaccination in the United Kingdom would lead to a reduction in the life-time cervical cancer risk of 73%, assuming 100% coverage of 12-year-old girls.13 Their result is also comparable with our result, taking into account that we assume an 85% coverage rate. In our study, we showed, in addition, that the impact of vaccination on the cervical cancer incidence is expected to be largest for women of 30–50 years of age. Preventing these early cancers will have a larger impact on life years saved than preventing cancers later in life. It is important to realize that the impact of vaccination depends on the starting age of screening, which is 30 years in The Netherlands. In countries like the United States where screening starts earlier, the impact of vaccination may be lower. In this regard, Kulasingam et al. quite tentatively suggest that cost-effectiveness is more likely when screening is delayed until later age when women are vaccinated. Finally, there are a number of model-based cost-effectiveness studies that present their results in formats that make direct comparison with our results more complicated. These studies report large differences in the ICERs, but most authors come to the conclusion that HPV16/18 vaccination is likely to be cost-effective.9–12, 14

As we developed a complex model with many different states of type-specific HPV infections, cervical lesions and cervical cancer, it is important to investigate the sensitivity of the model predictions for uncertainty in the disease parameters. Through a selection process, 39 different models were found that produced predictions in agreement with observed Dutch prevalences of HPV infection, CIN2/3 lesions and cervical cancer. These models were used to compute the cost-effectiveness of HPV vaccination in addition to screening. We found that the incremental cost-effectiveness ratio for our base-case vaccination scenario compared to screening only varied between € 11,000 and € 25,000 per QALY. The range lies within the range of ICERs reported in the international literature.9, 10, 14

We have shown that the influence of a decreasing screening compliance in vaccinated women has only a limited effect on the cost-effectiveness of HPV16/18 vaccination. However, we have not considered other effects of vaccination on future screening practice. The primary aim of screening in a vaccinated population is to detect (pre-) cancerous lesions caused by non-HPV16/18 infections. It has been argued that the positive predictive value of cytological screening decreases when the prevalence of cervical lesions and cervical cancer decreases.17, 52 This diminishes the effect of cervical screening on the cervical cancer incidence and may lead to a discussion about the cost-effectiveness of cervical screening in vaccinated women. To maintain the detection rate of cervical lesions after referral, screening algorithms must be carefully re-evaluated with regard to the choice of the screening instruments (cytology or HPV testing) and the length of the screening interval. Recent data indicate that extending the screening interval is feasible when using high-risk HPV testing as screening tool.16 Such alternative screening algorithms require formal modeling analyses.

Vaccination may also influence future screening practice by affecting screening compliance. It is not unrealistic that vaccinated women are less inclined to comply with cervical screening, if being vaccinated leads to a false sense of security. To keep screening compliance high, it is essential to inform women on the remaining risk of cervical cancer and the continuing need to be screened.

To conclude, our model calculations indicate that HPV16/18 vaccination could reduce the disease burden due to cervical cancer considerably if implemented in addition to the current screening program in The Netherlands. Estimations of cost-effectiveness lie around the current cost-effectiveness threshold in The Netherlands. The vaccine price and duration of protection induced by the vaccine are essential parameters in the cost-effectiveness analyses, i.e., vaccine price is linearly related to cost-effectiveness and shorter duration of protection leads to less favorable cost-effectiveness results. In case it is decided to include HPV vaccination in the national immunization program, the next important issue concerns the optimization of the cervical screening program. Furthermore, it is important to monitor the impact of these integrated programs to prevent cervical cancer.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

This study was funded by GlaxoSmithKline who had no role in the design of the study, the analysis and interpretation of the data or approval of the manuscript.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • 1
    Munoz N,Bosch FX,de Sanjose S,Herrero R,Castellsague X,Shah KV,Snijders PJ,Meijer CJ. Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med 2003; 348: 51827.
  • 2
    The Future II Study Group. Quadrivalent vaccine against human papillomavirus to prevent high-grade cervical lesions. N Engl J Med 2007; 356: 191527.
  • 3
    Harper DM,Franco EL,Wheeler CM,Moscicki AB,Romanowski B,Roteli-Martins CM,Jenkins D,Schuind A,Costa Clemens SA,Dubin G. Sustained efficacy up to 4.5 years of a bivalent L1 virus-like particle vaccine against human papillomavirus types 16 and 18: follow-up from a randomised control trial. Lancet 2006; 367: 124755.
  • 4
    Bosch FX,Manos MM,Munoz N,Sherman M,Jansen AM,Peto J,Schiffman MH,Moreno V,Kurman R,Shah KV. Prevalence of human papillomavirus in cervical cancer: a worldwide perspective.International biological study on cervical cancer (IBSCC) Study Group. J Natl Cancer Inst 1995; 87: 796802.
  • 5
    Clifford GM,Smith JS,Aguado T,Franceschi S. Comparison of HPV type distribution in high-grade cervical lesions and cervical cancer: a meta-analysis. Br J Cancer 2003; 89: 1015.
  • 6
    Ault KA. 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: 18618.
  • 7
    Paavonen J,Jenkins D,Bosch FX,Naud P,Salmeron J,Wheeler CM,Chow SN,Apter DL,Kitchener HC,Castellsague X,De Carvalho NS,Skinner SR, et al. Efficacy of a prophylactic adjuvanted bivalent L1 virus-like-particle vaccine against infection with human papillomavirus types 16 and 18 in young women: an interim analysis of a phase III double-blind, randomised controlled trial. Lancet 2007; 369: 216170.
  • 8
    Olsson SE,Villa LL,Costa RL,Petta CA,Andrade RP,Malm C,Iversen OE,Hoye J,Steinwall M,Riis-Johannessen G,Andersson-Ellstrom A,Elfgren K, et al. Induction of immune memory following administration of a prophylactic quadrivalent human papillomavirus (HPV) types 6/11/16/18 L1 virus-like particle (VLP) vaccine. Vaccine 2007; 25: 49319.
  • 9
    Elbasha EH,Dasbach EJ,Insinga RP. Model for assessing human papillomavirus vaccination strategies. Emerg Infect Dis 2007; 13: 2841.
  • 10
    Goldie SJ,Kohli M,Grima D,Weinstein MC,Wright TC,Bosch FX,Franco E. Projected clinical benefits and cost-effectiveness of a human papillomavirus 16/18 vaccine. J Natl Cancer Inst 2004; 96: 60415.
  • 11
    Kulasingam SL,Myers ER. Potential health and economic impact of adding a human papillomavirus vaccine to screening programs. JAMA 2003; 290: 7819.
  • 12
    Sanders GD,Taira AV. Cost-effectiveness of a potential vaccine for human papillomavirus. Emerg Infect Dis 2003; 9: 3748.
  • 13
    Kohli M,Ferko N,Martin A,Franco EL,Jenkins D,Gallivan S,Sherlaw-Johnson C,Drummond M. Estimating the long-term impact of a prophylactic human papillomavirus 16/18 vaccine on the burden of cervical cancer in the UK. Br J Cancer 2007; 96: 14350.
  • 14
    Brisson M,van de Velde N,de Wals P,Boily MC. The potential cost-effectiveness of prophylactic human papillomavirus vaccines in Canada. Vaccine 2007; 25: 5399408.
  • 15
    Bulkmans NW,Berkhof J,Bulk S,Bleeker MC,van Kemenade FJ,Rozendaal L,Snijders PJ,Meijer CJ. High-risk HPV type-specific clearance rates in cervical screening. Br J Cancer 2007; 96: 141924.
  • 16
    Bulkmans NW,Berkhof J,Rozendaal L,van Kemenade FJ,Boeke AJ,Bulk S,Voorhorst FJ,Verheijen RH,van Groningen K,Boon ME,Ruitinga W,van Ballegooijen M. 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: 176472.
  • 17
    Coupe VM,Berkhof J,Bulkmans NW,Snijders PJ,Meijer CJ. Age-dependent prevalence of 14 high-risk HPV types in The Netherlands: implications for prophylactic vaccination and screening. Br. J Cancer 2008; 98: 64651.
  • 18
    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: 26875.
  • 19
    Berkhof J,Bulkmans NW,Bleeker MC,Bulk S,Snijders PJ,Voorhorst FJ,Meijer CJ. Human papillomavirus type-specific 18-month risk of high-grade cervical intraepithelial neoplasia in women with a normal or borderline/mildly dyskaryotic smear. Cancer Epidemiol Biomarkers Prev 2006; 15: 126873.
  • 20
    Bulk S,Bulkmans NW,Berkhof J,Rozendaal L,Boeke AJ,Verheijen RH,Snijders PJ,Meijer CJ. Risk of high-grade cervical intra-epithelial neoplasia based on cytology and high-risk HPV testing at baseline and at 6-months. Int J Cancer 2007; 121: 3617.
  • 21
    Rozendaal L,Walboomers JM,van der Linden JC,Voorhorst FJ,Kenemans P,Helmerhorst TJ,van Ballegooijen M,Meijer CJ. PCR-based high-risk HPV test in cervical cancer screening gives objective risk assessment of women with cytomorphologically normal cervical smears. Int J Cancer 1996; 68: 7669.
  • 22
    Rozendaal L,Westerga J,van der Linden JC,Walboomers JM,Voorhorst FJ,Risse EK,Boon ME,Meijer CJ. PCR based high risk HPV testing is superior to neural network based screening for predicting incident CIN III in women with normal cytology and borderline changes. J Clin Pathol 2000; 53: 60611.
  • 23
    Gok M,Coupe VM,Berkhof J,Verheijen RH,Helmerhorst TJ,Hogewoning CJ,Snijders PJ,Meijer CJ. HPV16 and increased risk of recurrence after treatment for CIN. Gynecol Oncol 2007; 104: 2735.
  • 24
    Nobbenhuis MA,Walboomers JM,Helmerhorst TJ,Rozendaal L,Remmink AJ,Risse EK,van der Linden HC,Voorhorst FJ,Kenemans P,Meijer CJ. Relation of human papillomavirus status to cervical lesions and consequences for cervical-cancer screening: a prospective study. Lancet 1999; 354: 205.
  • 25
    Nobbenhuis MA,Helmerhorst TJ,van den Brule AJ,Rozendaal L,Voorhorst FJ,Bezemer PD,Verheijen RH,Meier CJ. Cytological regression and clearance of high-risk human papillomavirus in women with an abnormal cervical smear. Lancet 2001; 358: 17823.
  • 26
    Bulk S,Berkhof J,Bulkmans NW,Zielinski GD,Rozendaal L,van Kemenade FJ,Snijders PJ,Meijer CJ. Preferential risk of HPV16 for squamous cell carcinoma and of HPV18 for adenocarcinoma of the cervix compared to women with normal cytology in The Netherlands. Br J Cancer 2006; 94: 1715.
  • 27
    Bulk S,Berkhof J,Rozendaal L,Fransen Daalmeijer NC,Gok M,de Schipper FA,van Kemenade FJ,Snijders PJ,Meijer CJ. The contribution of HPV18 to cervical cancer is underestimated using high-grade CIN as a measure of screening efficiency. Br J Cancer 2007; 96: 123436.
  • 28
    Ketting BW. Surgical treatment of invasive carcinoma of the uterine cervix. Amsterdam: University of Amsterdam, 1981. 154.
  • 29
    van den Akker-van Marle ME,van BM,van Oortmarssen GJ,Boer R,Habbema JD. Cost-effectiveness of cervical cancer screening: comparison of screening policies. J Natl Cancer Inst 2002; 94: 193204.
  • 30
    Mandelblatt JS,Lawrence WF,Womack SM,Jacobson D,Yi B,Hwang YT,Gold K,Barter J,Shah K. Benefits and costs of using HPV testing to screen for cervical cancer. JAMA 2002; 287: 237281.
  • 31
    Berkhof J,de Bruijne MC,Zielinski GD,Bulkmans NW,Rozendaal L,Snijders PJ,Verheijen RH,Meijer CJ. Evaluation of cervical screening strategies with adjunct high-risk human papillomavirus testing for women with borderline or mild dyskaryosis. Int J Cancer 2006; 118: 175968.
  • 32
    van Ballegooijen M,Rebolj M,Essink-Bot ML,Meerding WJ,Berkers LM,Habbema JDF. De effected en kosten van het bevolkingsonderzoek naar baarmoederhalskanker in Nederland na de herstructurering. Rottterdam: Erasmus MC, 2006.
  • 33
    Nobbenhuis MA,Meijer CJ,van den Brule AJ,Rozendaal L,Voorhorst FJ,Risse EK,Verheijen RH,Helmerhorst TJ. Addition of high-risk HPV testing improves the current guidelines on follow-up after treatment for cervical intraepithelial neoplasia. Br J Cancer 2001; 84: 796801.
  • 34
    Zielinski GD,Snijders PJ,Rozendaal L,Voorhorst FJ,Runsink AP,de Schipper FA,Meijer CJ. High-risk HPV testing in women with borderline and mild dyskaryosis: long-term follow-up data and clinical relevance. J Pathol 2001; 195: 3006.
  • 35
    Sherman ME,Lorincz AT,Scott DR,Wacholder S,Castle PE,Glass AG,Mielzynska-Lohnas I,Rush BB,Schiffman M. Baseline cytology, human papillomavirus testing, and risk for cervical neoplasia: a 10-year cohort analysis. J Natl Cancer Inst 2003; 95: 4652.
  • 36
    Ostor AG. Natural history of cervical intraepithelial neoplasia: a critical review. Int J Gynecol Pathol 1993; 12: 18692.
  • 37
    Bos AB,van Ballegooijen M,van Oortmarssen GJ,van Marle ME,Habbema JD,Lynge E. Non-progression of cervical intraepithelial neoplasia estimated from population-screening data. Br J Cancer 1997; 75: 12430.
  • 38
    Briggs A,Sculpher M. An introduction to Markov modelling for economic evaluation. Pharmacoeconomics 1998; 13: 397409.
  • 39
    van Ballegooijen M,Boer R,van Oortmarssen GJ,Koopmanschap MA,Lubbe JTh,Habbema JD. Bevolkingsonderzoek op baarmoederhalskanker: leeftijdsgrenzen en intervallen. Een geactualiseerde kosten-effectiviteitsanalyse. Rotterdam: Instituut Maatschappelijke Gezondheidszorg, Erasmus Universiteit Rotterdam, 1993. 96.
  • 40
    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: 38893.
  • 41
    Garland SM,Hernandez-Avila M,Wheeler CM,Perez G,Harper DM,Leodolter S,Tang GW,Ferris DG,Steben M,Bryan J,Tadddeo FJ,Railkar R, et al. Quadrivalent vaccine against human papillomavirus to prevent anogenital diseases. N Engl J Med 2007; 356: 19281943.
  • 42
    Villa LL,Costa RL,Petta CA,Andrade RP,Paavonen J,Iversen OE,Olsson SE,Hoye J,Steinwall M,Riis-Johannesson G,Andersson-Ellstrom A,Elfgren K, et al. High sustained efficacy of a prophylactic quadrivalent human papillomavirus types 6/11/16/18 L1 virus-like particle vaccine through 5 years of follow-up. Br J Cancer 2006; 95: 145966.
  • 43
    van Ballegooijen M,Koopmanschap MA,Tjokrowardojo AJ,van Oortmarssen GJ. Care and costs for advanced cervical cancer. Eur J Cancer 1992; 28A: 17038.
  • 44
    Meijer CJ,Berkhof J,Heideman DA,Snijders PJ. Cervical cancer prevention: who should receive vaccination? Nat Clin Pract Oncol 2008; 5: 1213.
  • 45
    Liaw KL,Hildesheim A,Burk RD,Gravitt P,Wacholder S,Manos MM,Scott DR,Sherman ME,Kurman RJ,Glass AG,Anderson SM,Schiffman M. 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: 815.
  • 46
    Rousseau MC,Pereira JS,Prado JC,Villa LL,Rohan TE,Franco EL. Cervical coinfection with human papillomavirus (HPV) types as a predictor of acquisition and persistence of HPV infection. J Infect Dis 2001; 184: 150817.
  • 47
    Boot HJ,Wallenburg I,de Melker HE,Mangen MJ,Gerritsen AA,van der Maas NA,Berkhof J,Meijer CJ,Kimman TG. Assessing the introduction of universal human papillomavirus vaccination for preadolescent girls in The Netherlands. Vaccine 2007; 25: 624556.
  • 48
    van Veen M,Koedijk F,van den Broek I,Op de Coul E,de Boer I,van Sighem A,van der Sande MAB. Sexually transmitted infections in The Netherlands in 2006. RIVM rapport 210261003, Bilthoven, 2007.
  • 49
    van der Meijden WI,Notowicz A,Blog FB,Langley PC. A retrospective analysis of costs and patterns of treatment for external genital warts in The Netherlands. Clin Ther 2002; 24: 18396.
  • 50
    Giannini SL,Hanon E,Moris P,Van MM,Morel S,Dessy F,Fourneau MA,Colau B,Suzich J,Losonksy G,Martin MT,Dubin G, et al. Enhanced humoral and memory B cellular immunity using HPV16/18 L1 VLP vaccine formulated with the MPL/aluminium salt combination (AS04) compared to aluminium salt only. Vaccine 2006; 24(33–34): 593749.
  • 51
    Fraser C,Tomassini JE,Xi L,Golm G,Watson M,Giuliano AR,Barr E,Ault KA. Modeling the long-term antibody response of a human papillomavirus (HPV) virus-like particle (VLP) type 16 prophylactic vaccine. Vaccine 2007; 25: 432433.
  • 52
    Franco EL,Cuzick J,Hildesheim A,de Sanjose S. Chapter 20: issues in planning cervical cancer screening in the era of HPV vaccination. Vaccine 2006; 24 Suppl 3: S171S177.