Estimating long-term clinical effectiveness and cost-effectiveness of HPV 16/18 vaccine in China

Given its large population, China accounted for 11.7 % of the world’s cervical cancer (CC) cases and 11.3 % of the world’s CC deaths in 2012 [1]. Since 2009, a screening programme has been implemented in rural China targeting women aged 35 to 64 years, which has gradually expanded from screening 10 million women over a period of 3 years to screening 10 million women per year. Women in urban areas have access to CC screening on an opportunistic basis or through their employment. However, the mortality was reported to increase at an annual rate of 4.1% in young urban women aged 35 to 44 [2]. Furthermore, there are also wide disparities in CC incidence, mortality and screening accessibility between rural and urban areas in China [3].

Persistent HPV infection is the main cause of CC [4]. Seventy percent of all CC worldwide is caused by HPV 16/18 [5]; in China, HPV 16/18 infection accounts for 84.5 % of squamous cell carcinoma [6]. Prophylactic HPV vaccines may add to current efforts of screening in reducing CC burden in China. There are currently two vaccines available globally: a bivalent vaccine produced by GSK (Cervarix^®) and a quadrivalent vaccine produced by Merck (Gardasil^®). Both vaccines protect against HPV types 16 and 18 while the quadrivalent vaccine also protects against non-oncogenic types 6 and 11. Cervarix^®, a bivalent vaccine (GSK, Wavre, Belgium) has been widely used in over 100 countries through regional or national immunization programs to prevent HPV 16/18 infection and related diseases. Cervarix^® has shown high immunogenicity and safety, and induces a high degree of protection against HPV-16/18 infection and associated cervical lesions [7, 8]. To date, only Cervarix^® has just been approved by Chinese Food and Drug Administration in July 2016 after completation of phase III clinical trials using 3 doses of bivalent HPV vaccination (3DBV) in mainland China [9, 10].

Because the HPV vaccine is relatively costly [11], it has been subject to careful scrutiny. With the increasing importance of economic evaluation in priority setting for health [12], the cost-effectiveness of vaccination strategies should be considered before inclusion in national programmes, as recommended by the World Health Organization (WHO) position paper on HPV vaccines [13]. Mathematical models can help to estimate the long term effectiveness of HPV vaccination in parallel to clinical trials. Such models use a simplified description of the natural history leading to cervical cancer and provide a formal framework to synthesize information from various sources [14]. Modelling studies can integrate currently available clinical data with country-specific epidemiological data to evaluate the potential long-term impact of adding vaccination to screening [15].

Given the important resource implications of introducing the HPV vaccine in China, the financing of the HPV vaccination programme needs to be carefully considered to ensure the best use of resources compared to other priorities in China [16–18]. To date, there have been few comprehensive analyses evaluating the long-term impact of the HPV vaccine in China [19, 20]. Furthermore China is characterized by substantial differences in economic development and disease distribution between regions and therefore the diversity of economic, the ability to pay and screening influence should be considered to achieve the health care fairness. In such context, it is important to advance equity in China’s health system if implementing HPV vaccination in national vaccination program. We aimed to estimate the cost-effectiveness of HPV vaccination using Chinese specific parameters.

Under the circumstance that girls were regularly screened twice in their life time, considering the 3 % discount rate, 70 % screening and vaccination coverage and PAHO cost, HPV vaccination had an ICER of 11,365 CNY (US$ 1804) in rural and 6124 CNY (US$ 972) in urban, which is much lower than the national GDP 2013 (41,908 CNY, US$ 6652), which meant that HPV vaccination was very cost effective in this context.

Sensitivity analyses

The most influential factors on whether adding HPV vaccination to screening are cost-effective or not were evaluated in Fig. 2. The ICER varied largely when changing cost of vaccination (ICER ranged 142,110 CNY in rural and 132,128 CNY in urban), and discount (ICER ranged 49,272 CNY in rural and 42,885 CNY in urban) in one way sensitivity analysis. Other parameters, such as CC mortality, HPV infection, progression rate, vaccine efficacy, treatment cost of precancerous lesions, utility, age of screening and age at vaccination had little influence on the cost-effectiveness results. Shown from the one-way sensitivity analysis, when the CC mortality decreased, the HPV infection decreased, the vaccine efficacy against all cervical cancers irrespective of type decreased or the screening sensitivity increased, the vaccine cost should be reduced in order to remain the same ICER. In rural, the results showed that the most influential factors were cost of vaccination, discount rate, HPV infection rate, progression rate and vaccine efficacy (See Panel a). In urban, the most influential factors were cost of vaccination, discount rate, HPV infection rate, progression rate and cost of cancer (See Panel b). For screening coverage additionally, HPV vaccination was still very cost-effective (ICER = 14,761 CNY, US $ 2343 in rural and ICER = 9702 CNY, US$ 1540 in urban) with PAHO cost even in areas with 100 % screening coverage. In areas with the current low coverage (6.25 % [57] in rural and 21.5 % in urban [28]), the ICER would be 6463 CNY (US$ 1026) and 2135 CNY (US$ 339) in rural and urban areas, respectively. Under this circumstance, the 3DBV was estimated to be very cost-effective in both rural and urban.

The two-way sensitivity analysis showed the relationship between vaccine cost and three parameters identified in the one-way sensitivity (Fig. 3). Taken rural areas (Panel a, b, c) as an example, in panel a, the blue line with round dot was the results with 5 % discount; the blue line with square dot was the results with 3 % discount (base case value); the blue line with the diamond dot was the results with 0 % discount. The black lines showed the thresholds of cost effectiveness (three times national GDP) and very cost effectiveness (one time national GDP). Shown from panel a with 3 % discount, HPV vaccination was very cost-effective if the price was lower than 630 CNY (US$ 100) and it was cost-effective if the price was lower than 1700 CNY (US$ 270) in rural areas. With discount rate increased, the cost of HPV vaccine should be reduced to be cost-effective. On the other hand, with discount rate decreased, it was still cost-effective with the price of 1900 CNY (US$ 302). Similarly, for areas with higher CC mortality, HPV vaccination could be more cost-effective than in areas with lower CC mortality (See panel b). For areas with extremely lower screening coverage, HPV vaccination could be more cost-effective than in areas with higher screening coverage (See panel c). In urban, vaccination varied from very cost-effective to cost-effective with the national threshold when the vaccine cost ranged from 750 CNY (US$ 119) to 1900 CNY (US$ 302) in base case (Fig. 3, panel d, e, f).

Figure 4 showed the PSA in rural and urban. Each blue dot on the graph represented the relative discounted/undiscounted ICER of one out of the 10,000 simulations. X-axis represented the incremental QALYs and y-axis represented the incremental costs. The red dots represented the average value and the blue dots represented each replicates.

For rural setting, when comparing vaccination in addition to screening with screening only, with 3 % discount, 95.8 % of the replicates predicted to have greater lifetime costs and greater QALY benefit than the screening only scenario (Quadrant I of Fig. 4a). 4.2 % of dots fell into Quadrant II meant that vaccine combined with screening would cost less but gained more QALYs than screening only (Quadrant II in Fig. 4a). When undiscounted, 10.2 % of the replicates predicted to have greater lifetime costs and greater QALY benefit than the screening only in rural (Quadrant I of Fig. 4b). 89.8 % of dots fell into Quadrant II meant that vaccine combined with screening would cost less but gained more QALYs than screening only (Quadrant II in Fig. 4b). For urban setting, when comparing vaccination in addition to screening with screening only, with 3 % discount, 75.5 % of the replicates predicted to have greater lifetime costs and greater QALY benefit than the screening only scenario (Quadrant I of Fig. 4c). 24.5 % of dots fell into Quadrant II meant that vaccine combined with screening would cost less but gained more QALYs than screening only (Quadrant II in Fig. 4c). When undiscounted, 0.5 % of the replicates predicted to have greater lifetime costs and greater QALY benefit than the screening only in urban (Quadrant I of Fig. 4d). 99.5 % of dots fell into Quadrant II meant that vaccine combined with screening would cost less but gained more QALYs than screening only (Quadrant II in Fig. 4d).

The results showed that most of the replicates were in Quadrant I with 3 % discount and Quadrant II with 0 % discount, showing that vaccination in addition to screening had greater QALY benefit compared to screening only. All the replicates were less than the threshold of 3 times national GDP, that represented 100 % of replicates were cost-effectiveness in rural and urban respectively.

The ongoing screening programme with conventional Pap smear or visual inspection will inevitably miss many pre-cancer lesions due to the limitations of these two methods and the lack of health care workers [58]. Moreover, almost ten years delay of the HPV vaccine in China would cause more than 59 million Chinese girls aged 9-15 to lose their opportunities for vaccination [35] and put 206,000 girls at high risk for cervical cancer over the next 25 years [59]. HPV vaccine may add to current efforts of screening in reducing cervical cancer burden in China in the future.

In the current screening programme, considering the different situation in rural and urban areas, our study analysed two screening strategies respectively to reflect the difference in screening practice between rural and urban settings. Input parameters for rural and urban differed by HPV incidence, proportion of precancers treated, screening and treatment costs for precancerous lesions. The vaccine coverage was assumed the same (70 %) in rural and urban based on the perspective that government would include HPV vaccine in the national vaccination program as well as the success experiences of Hepatitis B vaccine in China. We also set the same methods for cancer treatment between rural and urban, given the realistic situation that most of the rural patients with diagnostically confirmed cervical cancer would seek for treatment in the urban hospitals due to no cancer therapeutic capacity in rural areas. The administration cost was US$ 9 per course, which was similar to previous published studies [18, 60]. The main findings showed that vaccination with 70 % coverage of 3DBV in 12-year-old girls in addition to screening could prevent 60 % more cervical cancer cases and deaths than screening only in China. In a case-control study in United States [61], the prevalence of HPV16/18 in CIN2+ lesions was demonstrated to decrease from 53.6 to 28.4 % among women received at least one dose of HPV vaccine (P _trend  < 0.001), which could lead to the reduction of cervical cancer in the future.

The perspective of health care payer was used in this study. Compared with societal perspective which considered the patient time, travel costs, willingness-to-pay and many other indirect parameters, the health care payer perspective could provide more information for the informing budget impact and policy decisions. As decisions on introduction of vaccines are often influenced by funding agencies, analysis such as budget impact assessments focusing on cost-saving, affordability and sustainability are relevant in resource-constrained settings [18, 62, 63]. The costs borne by providers (e.g., donors and governments) which separated by patients and their families could allow judgments to be made from the viewpoints of the various decision-makers. This is particularly important for middle and low income countries that may be required to fully-finance the cost of vaccines. Lieu [64] estimated (based on existing knowledge about the vaccine at the time) that from the health care payer perspective pneumococcal vaccination of healthy infants in the United States would result in savings if the vaccine cost US$ 18 or less per dose, but from the societal perspective, the vaccination programme would result in savings if the vaccine cost US$ 46 or less per dose. Analysts should therefore be cognizant that whilst a broader perspective that includes productivity losses (gains) will improve cost-effectiveness, it can also be used to justify higher vaccine cost, as it increases the break-even price per dose, i.e., the price at which the cost of the vaccination programme is exactly off-set by the savings due to vaccination [65].

In the sensitivity analysis, whether the results were cost-effective or not depended on the variability of the key input parameters. HPV vaccination was very cost-effective if the price was lower than 630 CNY (US$ 100) in rural and 750 CNY (US$ 119) in urban in two-way analysis. In order to be very cost-effective, the vaccine cost should be reduced in areas with lower cervical cancer mortality and higher screening coverage. Health care payers should compare the economic development level with the national threshold, CC mortality, HPV infection, HPV genotypes distribution, screening coverage, screening methods and other dominant parameters to decide the appropriate vaccination price. Global cost-effectiveness analysis informed by country-based evidence suggests that vaccinating pre-adolescent girls can be cost-effective, particularly in resource-constrained settings where alternative cervical cancer prevention and control measures often have limited coverage [66–69]. In the base case, we assumed the screening coverage as 70 % according to the previous research studies in China [50, 51], the coverage rate in real world setting would be much lower, for an instance, it was only 21.5 % in urban [28] and 6.25 % [57] in rural so far. For areas with low screening coverage and high cervical cancer burden, HPV vaccination could be more cost-effective and more urgent to be included into the vaccination plans.

Due to the diverse economic development of China, the different areas have different affordability profile. Since 2008, Panama has been the first country in Latin America and the Caribbean to provide the HPV vaccine free of charge to young adolescent girls [70]. After that, many developing countries have implemented the HPV vaccine [71, 72] into national programmes. In China, the government would strengthen the vaccination of HPV according to the China National Plan for NCD Prevention and Treatment [26]. However, the vaccine coverage and price would be the critical factors to be considered in the planning of vaccination programme.

In order to promote the implementation of HPV vaccine in China, multiple surveys have been done to gauge HPV vaccine awareness and knowledge. These multi-centre surveys in China showed that there was a high acceptability of HPV vaccine to prevent CC among Chinese women, but the price paid by individual should be considered carefully when the vaccines become available [73, 74]. The vaccine safety and efficacy would be the main concerns of HPV vaccine among Chinese [75]. Further proof of vaccine safety and efficacy and government subsidies combined with increased awareness could facilitate development and implementation of HPV vaccination in China.

Our analyses assumed the cost of 3 doses of HPV vaccine considering the registration clinical trial in China is for 3-dose vaccine regime. While the WHO position paper [13] on HPV vaccines proposed that the adoption of a 2-dose vaccine schedule could be recommended for girls aged 9–13 years. A 2-dose vaccine schedule might be more cost-effectiveness due to the non-inferior efficacy and less cost compared to 3-dose schedule.

Not only one-way sensitivity analysis, but also two-way analysis and probabilistic analysis were performed to certify the stability of the model. The main findings in our study were consistent with the previous published studies that evaluate the cost-effectiveness of HPV vaccination in China [19, 20]. We all found that combined HPV vaccination with screening would be more cost-effectiveness than only screening under a certain vaccine cost. The results in our study showed that adding vaccination to screening would reduce 60 % more cancer, while the other study showed a 44 % cancer reduction [20]. And the discrepancies between our study and others mainly occurred in the input parameters. The parameters in our Markov model were collected by multiple channels and validated by the expert panel, in which the experts came from different regions to fully capture the heterogeneity in disease burden, health systems and socio-economic development of China. The parameters were reliable and stable, which were estimated based on the pooled Chinese databases from multi-centre studies in rural and urban areas of 9 provinces across China [43]. Moreover, we used the most recently updated data from National Bureau of Statistics [23], China Cancer Registry [3] and Statistics of National Health and Family Planning Commission [58].

Limitations also exist in the methods. Firstly, the model did not capture the indirect protection resulting from the herd immunity caused by the reduction of circulation of the infective agent [76]. As a result, the benefits of the vaccination could be underestimated. Secondly, the vaccine efficacy in our model was assumed to be life-long, however,but the vaccine efficacy clinical trials have only lasted less than 10 years, so the duration of vaccine efficacy is unknown. Thirdly, the end study analysis of clinical registration trials of HPV vaccines in China are ongoing [9, 10] and no final data for the vaccine efficacy against CIN2/3+ irrespective of HPV types are available, so we adopted 93 % vaccine efficacy from PATRICIA trial [49]. The assumption of lifelong duration of vaccination and the high efficacy could overestimate the benefits of vaccination and result in more favourable ICERs. Fourthly, one of the uncertainties we have is whether HPV type replacement take place once vaccination against HPV-16/18 is widespread. The prevalence of HPV-16/18 falls to very low levels with vaccination. Other oncogenic HPV subtypes currently responsible for relatively few CC cases, might fill the niche left by HPV-16/18. To date this question could not be answered but available evidence suggests that niche competition is unlikely to happen [77]. Lastly, the present model only included cervical cancer and pre-cancer, irrespective of the vulvar cancer, vaginal cancer, anal and some proportion of oropharyngeal cancer cancers that the HPV vaccine may have efficacy in preventing [78–80]. If these diseases were included in the evaluation, the protection offered by HPV vaccination would be wider, and would lead to a lower ICER than the present analyses. These results are thus likely to provide a conservative estimate.

In both rural and urban areas, the vaccination cost and discounting are important factors determining the cost-effectiveness of HPV vaccination; policy makers in China should take these into account when making a decision on the introduction of HPV vaccine. In areas with a high burden of cervical cancer and limited screening activities, HPV vaccination should be prioritized. However, the vaccine cost needs to be reduced in order to make it very cost-effective and affordable as well, in particular in poverty areas with high disease burden.