Background: The ARTISTIC (A Randomised Trial In Screening To Improve Cytology) trial originally reported after two rounds of primary cervical screening with human papillomavirus (HPV). Extended follow-up of the randomised trial cohort through a third round could provide valuable insight into the duration of protection of a negative HPV test, which could allow extended screening intervals. If HPV primary screening is to be considered in the national programme, then determining its cost-effectiveness is key, and a detailed economic analysis using ARTISTIC data is needed.
Aims/objectives: (1) To determine the round 3 and cumulative rates of cervical intraepithelial neoplasia (CIN) grade 2 or worse (2+) and CIN grade 3 or worse (CIN3+) between the revealed and concealed arms of ARTISTIC after three screening rounds over 6 years. (2) To compare the cumulative incidence of CIN2+ over three screening rounds following negative screening cytology with that following negative baseline HPV. (3) To determine whether or not HPV screening could safely extend the screening interval from 3 to 6 years. (4) To study the potential clinical utility of an increased cut-off of 2 relative light unit/mean control (RLU/Co) for Hybrid Capture 2 (HC2) and HPV genotyping in primary cervical screening. (5) To determine the potential impact of HPV vaccination with Cervarix™ in terms of preventing abnormal cytology and CIN2+. (6) To determine the cost-effectiveness of HPV primary screening compared with current practice using cervical cytology in England.
Design: The ARTISTIC study cohort was recalled for a third round of screening 3 years after round 2 and 6 years following their enrolment to the study. Both arms of the original trial used a single protocol during round 3.
Setting: ARTISTIC study cohort undergoing cervical screening in primary care in Greater Manchester, UK.
Participants: Between July 2007 and September 2009, 8873 women participated in round 3; 6337 had been screened in round 2 and 2536 had not been screened since round 1.
Interventions: All women underwent liquid-based cytology and HPV testing and genotyping. Colposcopy was offered to women with moderate dyskaryosis or worse and with HPV-positive mild dyskaryosis/borderline changes. Women with negative cytology or HPV-negative mild dyskaryosis/borderline changes were returned to routine recall.
Main outcome measures: Principal outcomes were cumulative rates of CIN2+ over three screening rounds by cytology and HPV status at entry; HPV type specific rates of CIN2+; effect of age on outcomes correlated with cytology and HPV status; comparison of HC2 cut-off RLU/Co of both 1 and 2; and cost-effectiveness of HPV primary screening.
Results: The median duration of follow-up was 72.7 months in round 3. Over the three screening rounds, there was no significant difference in CIN2+ [odds ratio (OR): 1.06, 95% confidence interval (CI) 0.89 to 1.26, p = 0.5)] or CIN3+ (OR: 0.90, 95% CI 0.72 to 1.14, p = 0.4) rates between the trial arms (revealed vs. concealed). Overall, 16% of women were HC2 positive at entry, decreasing from 40% in women aged 20-24 years to around 7% in women aged over 50 years. Abnormal cytology rates at entry were 13% for borderline+ and 2% for moderate+ cytology. Following positive cytology at entry, the cumulative rate of CIN2+ was 20.5%, and was 20.1% following a HPV-positive result at baseline. The cumulative CIN2+ rate for women who were HPV negative at baseline was only 0.87% (95% CI 0.70% to 1.06%) after three rounds of screening, significantly lower than that for women with negative cytology, which was 1.41% (95% CI 1.19% to 1.65%). Women who were HPV negative at baseline had similar protection from CIN2+ after 6 years as women who were cytology negative at baseline after 3 years. Women who were HPV positive/cytology negative at baseline had a cumulative CIN2+ rate at 6 years of 7.7%, significantly higher than that for women who were cytology positive/HPV negative (3.2%). Women who were HPV type 16 positive at baseline had a cumulative CIN2+ rate over three rounds of 43.6% compared with 20.1% for any HPV-positive test. Using a HC2 cut-off of RLU/Co ≥ 2 would maintain acceptable sensitivity and result in 16% fewer HPV-positive results. Typing data suggested that around 55-60% of high-grade cytology and CIN2+, but less than 25% of low-grade cytology, would be prevented by HPV vaccine given current rates of coverage in the UK national programme. For the cost-effectiveness analysis, most of the primary HPV strategies examined where HPV was used as the sole primary test were cost saving in both unvaccinated and vaccinated cohorts under baseline cost assumptions, with a 7-18% reduction in annual screening-associated costs in unvaccinated cohorts and a 9-22% reduction for vaccinated cohorts. Utilising partial genotyping at the primary screening stage to identify women with HPV 16/18 and referring them to colposcopy was the most effective strategy (barring co-testing, which is significantly more costly than any other strategies considered), resulting in 83 additional life-years per 100,000 women for unvaccinated women when compared with current practice, and similar life-years saved compared with current practice for vaccinated women. In unvaccinated cohorts, however, this genotyping strategy is predicted to result in a 20% increase in the number of colposcopies performed in England, although in vaccinated cohorts the number of colposcopy referrals was predicted to be lower than in current practice. For all strategies in which HPV is used as the sole primary screening test, decreasing the follow-up interval for intermediate-risk women from 24 to 12 months increased the overall effectiveness of primary HPV screening. In exploratory analysis, strategies for which cytology screening was retained until either age 30 or 35 years, and for which HPV testing was used at older ages, were predicted to be of higher costs and intermediate effectiveness than those associated with full implementation of primary HPV screening from age 25 years. However, this finding should be interpreted with caution as it depends on assumptions made about screening behaviour and compliance with recommendations at the 'switch over' point.
Conclusions: HPV testing as an initial screen was significantly more protective over three rounds (6 years) than the current practice of cytology and the use of primary HPV screening could allow a safe lengthening of the screening interval. A substantial decrease in high-grade cytology and CIN2+ can be expected as a consequence of the HPV vaccination programme. A HC2 cut-off of 2RLU/Co instead of the manufacturer's recommended cut-off of 1 would be clinically beneficial in terms of an optimal balance between sensitivity and specificity. Modelled analysis predicts that primary HPV screening would be both more effective and cost saving compared with current practice with cervical cytology for a number of potential strategies in both unvaccinated and vaccinated cohorts. Compliance with surveillance and optimal management of HPV-positive/cytology-negative women after primary HPV screening is of key importance. Limitations of the economic investigation included the need to make assumptions around compliance with screening attendance and follow-up for longer screening intervals in the future, assumptions regarding maintenance of current uptake vaccination in the future, and assumptions regarding the stability of cost of HPV and cytology tests in the future. Detailed sensitivity analysis across a range of possible assumptions was conducted to address these issues. This study and the economic evaluation lend support to convert from cytology to HPV-based screening. Future work should include researching (i) the attitudes of women who test HPV positive/cytology negative, (ii) the value of complementary biomarkers and (iii) activities relevant to primary HPV screening in unvaccinated and vaccinated populations from the point of view of QALY assessment.
Study registration: Current Controlled Trials ISRCTN25417821.