Prediction of cervical cancer incidence in England, UK, up to 2040, under four scenarios: a modelling study

Abstract

Background: In the next 25 years, the epidemiology of cervical cancer in England, UK, will change: human papillomavirus (HPV) screening will be the primary test for cervical cancer. Additionally, the proportion of women screened regularly is decreasing and women who received the HPV vaccine are due to attend screening for the first time. Therefore, we aimed to estimate how vaccination against HPV, changes to the screening test, and falling screening coverage will affect cervical cancer incidence in England up to 2040.

Methods: We did a data modelling study that combined results from population modelling of incidence trends, observable data from the individual level with use of a generalised linear model, and microsimulation of unobservable disease states. We estimated age-specific absolute risks of cervical cancer in the absence of screening (derived from individual level data). We used an age period cohort model to estimate birth cohort effects. We multiplied the absolute risks by the age cohort effects to provide absolute risks of cervical cancer for unscreened women in different birth cohorts. We obtained relative risks (RRs) of cervical cancer by screening history (never screened, regularly screened, or lapsed attender) using data from a population-based case-control study for unvaccinated women, and using a microsimulation model for vaccinated women. RRs of primary HPV screening were relative to cytology. We used the proportion of women in each 5-year age group (25-29 years to 75-79 years) and 5-year period (2016-20 to 2036-40) who have a combination of screening and vaccination history, and weighted to estimate the population incidence. The primary outcome was the number of cases and rates per 100 000 women under four scenarios: no changes to current screening coverage or vaccine uptake and HPV primary testing from 2019 (status quo), changing the year in which HPV primary testing is introduced, introduction of the nine-valent vaccine, and changes to cervical screening coverage.

Findings: The status quo scenario estimated that the peak age of cancer diagnosis will shift from the ages of 25-29 years in 2011-15 to 55-59 years in 2036-40. Unvaccinated women born between 1975 and 1990 were predicted to have a relatively high risk of cervical cancer throughout their lives. Introduction of primary HPV screening from 2019 could reduce age-standardised rates of cervical cancer at ages 25-64 years by 19%, from 15·1 in 2016 to 12·2 per 100 000 women as soon as 2028. Vaccination against HPV types 16 and 18 (HPV 16/18) could see cervical cancer rates in women aged 25-29 years decrease by 55% (from 20·9 in 2011-15 to 9·5 per 100 000 women by 2036-40), and introduction of nine-valent vaccination from 2019 compared with continuing vaccination against HPV 16/18 will reduce rates by a further 36% (from 9·5 to 6·1 per 100 000 women) by 2036-40. Women born before 1991 will not benefit directly from vaccination; therefore, despite vaccination and primary HPV screening with current screening coverage, European age-standardised rates of cervical cancer at ages 25-79 years will decrease by only 10% (from 12·8 in 2011-15 to 11·5 per 100 000 women in 2036-40). If screening coverage fell to 50%, European age-standardised rates could increase by 27% (from 12·8 to 16·3 per 100 000 by 2036-40).

Interpretation: Going forward, focus should be placed on scenarios that offer less intensive screening for vaccinated women and more on increasing coverage and incorporation of new technologies to enhance current cervical screening among unvaccinated women.

Funding: Jo's Cervical Cancer Trust and Cancer Research UK.