Predicting cohort-specific cervical cancer incidence from population-based HPV prevalence surveys: a worldwide study

Background Predictions of cervical cancer burden and impact of control measures are often modelled from HPV prevalence. However, predictions could be improved by data on time between prevalent HPV detection and cervical cancer occurrence. Methods Based upon high-risk (HR) HPV prevalence and cervical cancer incidence in the same birth cohorts from 17 worldwide locations, and informed by individual-level data on age at HR HPV detection and on sexual debut, we built a mixed model to predict cervical cancer incidence up to 14 years following prevalent HR HPV detection. Findings Cervical cancer incidence increased significantly during the 14 years following HR HPV detection in women <35 years, e.g. from 0·02 (95% CI 0·003–0·06) per 1000 within 1 year to 2·8 (1·2–6·5) at 14 years for unscreened women, but remained relatively constant following prevalent HR HPV detection above 35 years, e.g. from 5·4 (2·5–11) per 1000 within 1 year to 6·4 (2·4–17·1) at 14 years for unscreened HR HPV positive women aged 45–54 years. Age at sexual debut was a significant modifier of cervical cancer incidence in HR HPV positive women aged <25, but less so at older ages, whereas screening was a modifier in women ≥35 years. Lastly, we predicted annual number and incidence of cervical cancer in ten additional IARC HPV prevalence survey locations without representative cancer incidence data. Interpretation These findings can inform cervical cancer control programmes, particularly in settings without cancer registries, as they allow prediction of future cervical cancer burden from population-based surveys of HPV prevalence. Funding Bill & Melinda Gates Foundation; Canadian Institutes of Health Research.