In this chapter, we summarise findings from recent cost-effectiveness analyses of screening for cervical cancer and ovarian cancer. We begin with a brief summary of key issues that affect the cost-effectiveness of screening, including disease burden, and availability and type of screening tests. For cervical cancer, we discuss the potential effect of human papilloma virus vaccines on screening. Outstanding epidemiological and cost-effectiveness issues are included. For cervical cancer, this includes incorporating the long-term effect of treatment (including adverse birth outcomes in treated women who are of reproductive age) into cost-effectiveness models using newly available trial data to identify the best strategy for incorporating human papilloma virus tests. A second issue is the need for additional data on human papilloma virus vaccines, such as effectiveness of reduced cancer incidence and mortality, effectiveness in previously exposed women and coverage. Definitive data on these parameters will allow us to update model-based analyses to include more realistic estimates, and also potentially dramatically alter our approach to screening. For ovarian cancer, outstanding issues include confirming within the context of a trial that screening is effective for reducing mortality and incorporating tests with high specificity into screening into screening algorithms for ovarian cancer.
Introduction
In this chapter, we will discuss findings to date from health-economic evaluations of screening strategies for cervical and ovarian cancer prevention. In particular, we focus on recently published cost-effectiveness analyses or reviews. The chapter is organised into two sections: the first focuses on cervical cancer; the second on ovarian cancer. Preceding each section are short summaries of key aspects of screening programmes that influence the cost-effectiveness of screening for these two cancers.
Criteria for implementing a screening programme for any disease, including cancer, are based on the following (adapted from World Health Organization, 1968) : (1) the condition sought should be an important health problem; (2) accepted treatments should be available for women with recognised disease; (3) facilities for diagnosis and treatment should be available; (4) there should be a recognisable latent or early symptomatic stage; (5) a suitable test or examination should be available; (6) the test should be acceptable to the population; (7) the natural history of the condition, including development from latent to declared disease, should be adequately understood; (8) a policy should be agreed on whom to treat as patients; (9) the cost of case-finding (including diagnosis and treatment of patients diagnosed) should be economically balanced in relation to possible expenditure on medical care as a whole; and (10) case-finding should be a continuing process and not a ‘once and for all’ project.
Of these, aspects that have been key to influencing the cost-effectiveness of screening for cervical and ovarian cancer are the burden and natural history of disease, the availability of accurate screening tests, effectiveness of treatment, especially at the early stages of disease, and the costs of implementing and maintaining a screening programme.
Burden and natural history of cervical cancer
Cervical cancer is the third most common cancer-related cause of death in women worldwide. Each year, about 500,000 new cases of cervical cancer are diagnosed, with over 80% of cases occurring in developing countries ( Fig. 1 ). Infection with human papilloma virus (HPV) is considered to be necessary for the development of cervical cancer. Human papilloma virus is a common infection that is acquired through sexual activity. Although most women infected with HPV eventually clear their infections, some go on to develop pre-cancerous changes termed ‘cervical intraepithelial neoplasia’ (CIN). Cervical intraepithelial neoplasia is categorised by the proportion of cervical epithelial cells showing abnormalities and their degree of atypia. It is sub-divided into CIN 1 (mild dysplasia), CIN 2 (moderate dysplasia) and CIN 3 (severe dysplasia). Although evidence supports the poor reproducibility of CIN 1 and 2 in particular, CIN continues to be used to categorise cervical pre-cancer. If left untreated, CIN 3 lesions can progress to cervical cancer. As the time between CIN and the development of invasive cervical cancer can be decades-long, pre-invasive lesions can potentially be detected through screening.
Burden and natural history of cervical cancer
Cervical cancer is the third most common cancer-related cause of death in women worldwide. Each year, about 500,000 new cases of cervical cancer are diagnosed, with over 80% of cases occurring in developing countries ( Fig. 1 ). Infection with human papilloma virus (HPV) is considered to be necessary for the development of cervical cancer. Human papilloma virus is a common infection that is acquired through sexual activity. Although most women infected with HPV eventually clear their infections, some go on to develop pre-cancerous changes termed ‘cervical intraepithelial neoplasia’ (CIN). Cervical intraepithelial neoplasia is categorised by the proportion of cervical epithelial cells showing abnormalities and their degree of atypia. It is sub-divided into CIN 1 (mild dysplasia), CIN 2 (moderate dysplasia) and CIN 3 (severe dysplasia). Although evidence supports the poor reproducibility of CIN 1 and 2 in particular, CIN continues to be used to categorise cervical pre-cancer. If left untreated, CIN 3 lesions can progress to cervical cancer. As the time between CIN and the development of invasive cervical cancer can be decades-long, pre-invasive lesions can potentially be detected through screening.
Screening for cervical cancer
Although a number of observational studies have examined the effect of screening (whether HPV or cytology) on CIN, until recently no randomised-controlled trial (RCT) has shown that screening (and treatment of screen detected lesions) reduces cancer incidence and mortality. Evidence for the effect of screening is based on observational studies that show a dramatic decline in cervical cancer incidence after the introduction of wide-spread cytology-based screening. In 2009, an RCT conducted in India compared screening with cytology or HPV tests to no screening, and confirmed that screening was indeed effective in reducing incidence and mortality.
Screening programmes to detect CIN and early stage cervical cancer vary by country and setting, and are based on conventional Pap smears, liquid-based cytology (LBC), HPV DNA tests, or both. Although early studies of LBC suggested an improved sensitivity compared with conventional cytology, recent results from meta-analyses and RCTs confirm no statistically significant difference between the two for the detection of CIN 2 or greater (CIN2+).
Over 40 types of HPV are sexually transmitted, but at least 15 are commonly found in cervical cancer and are termed high risk or oncogenic. Human papilloma virus screening tests, which can detect the presence of some of these high-risk types, include the hybrid capture 2 test (Qiagen, Gaithersburg, MD, USA) and the Cervista ® HPV HR (Hologic Inc, Madison, WI, USA). Other tests designed specifically to detect the presence of HPV 16 and 18 (two HPV types that are associated with about 70% of cervical cancers worldwide) include Roche’s cobas ® 4 (F. Hoffmann-La Roche Ltd, Basel, Switzerland) and Hologic Inc.’s Cervista ® HPV 16/18. A recent meta-analysis concluded that, compared with the Pap smear, the hybrid capture 2 test detects a much higher percentage of CIN2+ (96.1% compared with 53.0%), but has a lower specificity (90.7% compared with 96.3%). Given the lower specificity, recommendations are for HPV DNA tests to be used as either a triage test for low-grade cytological abnormalities or as a primary screening test in combination with cytology, with use restricted to women aged 30 years or older. The reason for restricting by age is that, although HPV prevalence declines with age, the prevalence of CIN 2–3 increases. Therefore, restricting HPV testing reduces the likelihood of false–positive results.
Several trials that are currently ongoing or recently completed have compared the performance of HPV in combination with cytology to that of cytology alone. To date, these trials have shown that the addition of HPV tests to cytology-based programmes increases detection of CIN 2+, especially during the first round of screening with HPV. This gain in disease detection, however, comes at the expense of additional false–positive test results. Women with negative HPV test results have been shown to have a much lower risk of subsequent disease; this suggests that a way to include HPV DNA tests efficiently is to screen women who are HPV negative less frequently than women with normal cytology results. To that end, updated screening guidelines, such as those from the American Society of Colposcopy and Cervical Pathology, which include HPV DNA testing, recommend longer screening intervals for women who are HPV DNA negative; the optimal length of interval is still under study.
Treatment for cervical intraepithelial neoplasia
Treatment options for CIN include ablative techniques, cone biopsy, loop electrosurgical excision procedure and hysterectomy. Contrary to the notion that screening and treatment reduce subsequent risk in treated women, literature reviews and a recent cohort study suggest that those who have been treated for CIN may be at an increased risk of CIN and cancer after treatment, and may require closer surveillance than the general population. Soutter et al. in a review of studies on outcomes in women treated for CIN, reported a higher risk of invasive cancer, but lower risk of CIN in women. As it is impossible to distinguish whether invasive cancer is caused by a recurrence or a new occurrence of disease, the investigators concluded that increased and sustained screening in treated women is needed to ensure that treatment does indeed reduce the risk of subsequent disease. Similar results were also reported in a recently published retrospective cohort study from the British Columbia Cancer agency. The cytology database was linked with the cancer and death registries, and the long-term effectiveness of treatment for CIN was examined. It was found that, although treatment resulted in long-term (defined as more than 6 years) rates of CIN 2–3 that were comparable to those in women with no CIN, the incidence of CIN 2–3 was higher in women treated over the short term than in untreated women. The overall incidence of invasive cancer (per 100,000 woman-years) was also higher in the CIN cohort (37 invasive cancers, 95% confidence interval [CI] = 30.6 to 42.5 cancers) than in the comparison cohort (six cancers, 95% CI = 4.3 to 7.7 cancers), with the highest incidence associated with those treated with cryotherapy (adjusted odds ratio for invasive cancer = 2.98, 95% CI = 2.09 to 4.60).
Adding to the concerns about recurrent or incident disease in women who have been previously treated for CIN are results from cohort studies and meta-analyses that show an increased risk of adverse birth outcomes for women of childbearing age who are treated. Of the different treatment options available, cold-knife conisation, in particular, has been shown to be associated with a significantly increased risk of adverse birth outcomes. A systematic review and meta-analysis by Arbyn et al. showed that cold-knife conisation was associated with increased perinatal mortality (relative risk 2.87, 95% CI 1.42 to 5.81) and a significantly higher risk of severe preterm delivery (2.78, 1.72 to 4.51), extreme preterm delivery (5.33, 1.63 to 17.40), and low birth weight of less than 2000 g (2.86, 1.37 to 5.97).
Human papilloma virus vaccines
Screening has been successful in reducing cancer incidence and mortality. This has primarily been accomplished by screening women frequently and treating those with confirmed disease; estimates from the USA suggest that more than $4 billion is spent annually in providing cervical cancer screening and follow up. Other shortcomings, in addition to the high costs of maintaining a successful screening programme, exist. As discussed briefly, studies that provide the most accurate estimates of screening test sensitivity suggest that conventional cytology has a sensitivity close to 50% and that the sensitivity of LBC is not significantly different from that of conventional cytology. As stated previously, HPV DNA testing has improved sensitivity for detecting high-grade pre-cancer, but decreased specificity compared with cytology. Another limitation of screening is that, in countries such as Mexico, which have implemented cytology-based screening, quality control and expanding access remain issues. Importantly, cervical cancer remains a leading cause of cancer-related deaths among women in countries that have not implemented cervical cancer screening, whether due to economic or other issues. As such, other non-screening-based options for cervical cancer prevention are needed.
Two vaccines, Gardasil ® developed by Merck (Whitehouse Station, New Jersey, USA) and Cervarix ® , developed by GlaxoSmithKline (Brentford, Middlesex, UK) targeted at HPV types 16 and 18 account for about 70% of cervical cancers worldwide, and are currently available. Three systematic reviews of the results of these trials show high efficacy for both vaccines. The most recent pooled data from trials for both Cervarix ® and Gardasil ® , and showed a significantly lower risk of CIN 2+ in vaccinated women compared with unvaccinated women. Reductions, however, differed by HPV type (fixed-effect relative risk 0.04 [0.01 to 0.11] and 0.10 [0.03 to 0.38] for HPV-16 and HPV 18-related CIN2+ in the per protocol populations, and by whether vaccinated women had evidence of infection with a vaccine-relevant HPV type. The per protocol estimates suggest that, in unvaccinated women who have not been exposed to the vaccine types, the vaccines will be highly effective in over 90% of women. The intention-to-treat analyses, which included women who were infected with HPV at baseline, confirm that the efficacy will be lower in women who have already been exposed to the relevant HPV types (i.e. are sexually active), don’t receive all three doses of the vaccine per the stated protocol, or both. A recently published analysis of vaccine efficacy among women who had detectable serological evidence of vaccine-HPV infection in the past but no DNA evidence of active infection at enrollment showed that the vaccine provides high efficacy against CIN2+ associated with the same vaccine HPV type. These data suggest that women of older reproductive age may still benefit from prophylactic vaccination; what remains unknown is what proportion of women in the general population may benefit and whether this proportion is high enough to justify increasing the recommended age for vaccination in countries that have restricted the recommendation under to age 26 years. A related issue concerning efficacy is cross-protection for non-vaccine HPV types, in particular 31, 33, 45, 52, 58. The degree of cross-protection differs by type. On the basis of data from trials of both Cervarix ® and Gardasil ® ), Lu et al. reported an approximate 40% reduction in CIN 2+ related to these HPV types in the per protocol group and a 20% reduction for the intention to treat groups. This suggests that the vaccines will also provide additional protection (albeit lower) for the additional cancer attributable to these types. Finally, although data have yet to be published, an octavalent vaccine developed by Merck and targeted at HPV types, which accounts for over 90% of disease, is currently being studied in RCTs.
A key parameter (discussed below) for these vaccines is the duration of vaccine efficacy. Although the logarithmically higher antibody titers resulting from vaccine suggest long and sustained immunity in vaccinated women, data confirming this and, more importantly, confirming that this sustained immunity will result in reductions in cervical cancer are pivotal but remain lacking. Taken together, however, these results suggest that vaccines have the potential to considerably change current approaches to screening.
Summary of cost-effectiveness studies
Screening only
Until the mid-2000s, most cost-effectiveness analyses examined questions related to screening only. These questions focused on the use of newly available screening technologies, in particular LBC and HPV DNA testing, for primary screening and for cytology results of atypical squamous cells of undetermined significance. Earlier cost-effectiveness studies suggested that switching to LBC would be effective, cost-effective, or both; however, questions remained about the true gain in sensitivity compared with conventional cytology. In particular, a health technology assessment by Karnon et al. concluded that more definitive data on LBC test accuracy was needed. Since then, data from RCTs and meta-analyses have confirmed no significant difference between the two tests in detection of CIN 2+. These findings highlight the need for caution when interpreting cost-effectiveness studies where data on test performance are of limited quality. A review of cost-effectiveness studies of HPV DNA testing in addition to cytology shows that, across a range of analyses, screening with HPV DNA tests (in particular HC2) is potentially cost-effective if it allows for a change in screening interval in older women (defined as either age 30 or 35 years depending on the setting). For instance, an analysis by Goldie et al. showed that adding HPV DNA testing to cytology (co-testing) or for atypical squamous cells of undetermined significance would be cheaper and result in increased life-expectancy if conducted at 2–3-year intervals compared with cytology-only screening conducted at the same or more frequent intervals. In addition, these analyses repeatedly show that annual cytology-based screening is associated with high costs and many more procedures for relatively small gains in life-expectancy (on the order of hours per woman). More recent cost-effectiveness studies, based on ongoing or recently completed trials, also suggest that, of the different strategies available for combining HPV and cytology, a strategy of HPV testing followed by cytology for women with positive HPV test results may be a cost-effective way to include HPV DNA testing in screening for older women.
In resource-limited settings, questions have focused on what type of screening test to use (e.g. HPV DNA, cytology, direct visual inspection, or all) and how to maximise disease detection and treatment while minimising visit burden. These analyses have identified screen and treat (at either the same or a different visit) as the most cost-effective options for maximising reductions in cancer. In particular, cervical-cancer screening strategies incorporating visual inspection of the cervix with acetic acid or DNA testing for HPV in one or two clinical visits are cost-effective alternatives to conventional three-visit, cytology-based screening programmes in resource-poor settings. Although treatment failure with cryotherapy was modelled over 1 year, cryotherapy has been shown to be associated with a higher risk of recurrent CIN and cancer over an extended period of time (as long as 6 years). This highlights the need for careful consideration of the benefits and harms associated with screen-and-treat strategies, and perhaps the need to revisit the issue of optimal screening strategies for resource-poor countries.
Vaccination and screening
Model form
The availability of efficacious vaccines to prevent HPV infection and disease offers the potential to prevent invasive cervical cancer, cervical pre-cancer, genital warts and other anogenital cancers. Because the epidemiology and natural history of HPV infection and cervical cancer are complex, how these vaccines will be used (who will be vaccinated and when) is a non-trivial question. In addition, effectiveness and cost-effectiveness of the vaccines vary depending upon the role of organised screening, the vaccination strategy, and the cost of the immunisation programme. As the vaccine trials use high-grade dysplasia rather than cancer as an outcome, cost-effectiveness modelling has been used to project the effect of interrupting infection with HPV on cervical cancer incidence and mortality.
Two types of models have been used to explore HPV vaccine effectiveness: cohort models and dynamic transmission models, with a third category, hybrid models, using a combined approach. Cohort or health-state transition models are the most common type of model used for examining screening strategies, and are generally probabilistic and linear; they simulate the progression of HPV disease for a single population cohort over its expected lifetime. Dynamic models to date have been generally deterministic and non-linear: they track populations over time taking into account births and deaths. Importantly, dynamic transmission models account for the rate of infection, which is dependent on patterns of sexual behaviour, and the distribution of infection in the population. The strengths of dynamic models are that they can be used to determine the herd immunity effects of vaccination programmes, explore the relative value of vaccinating boys in addition to girls, and explore the effect of sexual mixing patterns on age at vaccination. For HPV, dynamic models have been used to determine the reduction in HPV incidence relative to the proportion of the population is vaccinated. For example, in a model of HPV transmission developed by Barnabas et al. vaccinating 80% of women alone before sexual debut reduced the incidence of HPV in women aged between 15 and 19 years by 92%.
A limitation of dynamic models is that the complexity of modelling HPV as a sexually transmitted infection has been achieved at the expense of greatly simplifying modelling of screening. For screening programmes (i.e. those in the USA) that use complicated triage algorithms, this may mean an underestimation of the benefits of screening. To address this, hybrid models have been developed that use both approaches. Hybrid models generate HPV incidence under different vaccination scenarios from a dynamic model and then input these into a cohort model to determine the effect of herd immunity on reductions in cancer incidence and mortality while accounting for complex screening programmes.
Cost-effectiveness of screening and vaccination
A number of reviews of different cost-effectiveness models have been published over the past few years. Across a variety of settings, and taking into account variations in model structure as described previously, these analyses suggest that HPV vaccines have the potential to considerably reduce the burden of CIN and cervical cancer in settings both with and without screening. These findings are, however, sensitive to a number of assumptions, key among them a long duration of vaccine efficacy and high vaccine coverage. Although vaccine duration will remain an unknown for the foreseeable future, coverage has lagged modelled estimates, suggesting the need to update analyses based on lower than modelled coverage. Vaccine price has also consistently been shown to affect the cost-effectiveness of adding vaccination to screening. Indeed, if HPV vaccines are priced below certain thresholds for different countries, HPV vaccination could potentially be cost-saving compared with doing nothing. Finally, as would be expected on the basis of trial results, vaccination of girls only before onset of sexual activity, compared with vaccination of boys, girls and catch-up vaccination, has been shown to have the most attractive cost-effectiveness profile.
Screening in the era of human papilloma virus vaccines
The issue of whether and how screening should change in the era of HPV vaccines is complex, and will depend on a number of factors that are still unknown. These include, but are not limited to, the performance of cytology and HPV tests (i.e. hybrid capture 2), which are not type specific, reductions in CIN and cancer (the latter is an outcome that may well take decades to determine), and whether vaccination will affect screening behaviour, in particular screening participation. Under the assumption that vaccines will markedly reduce cancer incidence and mortality, potentially cost-effective approaches to screening vaccinated cohorts of women include strategies that use a less frequent screening interval, a delayed age of first screening, or use of a strategy based on HPV DNA testing followed by cytology. These approaches point to the need to reconsider how to screen for a low-incidence cancer, which cervical cancer may well become in the era of HPV vaccines. The potential effect of a bivalent and octavalent vaccine on cancer incidence is shown in Fig. 1 . Under certain conditions, cervical cancer incidence may well mirror that of a low incident cancer, such as ovarian cancer.
In the next section, we discuss issues around the cost-effectiveness of screening for ovarian cancer, and particularly those that may well be pertinent to cervical cancer in the coming decades. Given the current uncertainty around some key HPV vaccine parameters, however, no changes to screening are anticipated for the foreseeable future.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
