Impact of cervical screening by human papillomavirus genotype: Population-based estimations

BACKGROUND: Cervical screening programs use testing for human papillomavirus (HPV) genotypes. Different HPV types differ greatly in prevalence and oncogenicity. We estimated the impact of cervical screening and follow-up for each HPV type. METHODS AND FINDINGS: For each type of HPV, we calculated the number of women needed to screen (NNS) and number of women needing follow-up (NNF) to detect or prevent one cervical cancer case, using the following individual level input data (i) screening and cancer data for all women aged 25 to 80 years, resident in Sweden during 2004 to 2011 (N = 3,568,938); (ii) HPV type-specific prevalences and screening histories among women with cervical cancer in Sweden in 2002 to 2011(N = 4,254); (iii) HPV 16/18/other HPV prevalences in the population-based HPV screening program (N = 656,607); and (iv) exact HPV genotyping in a population-based cohort (n = 12,527). Historical screening attendance was associated with a 72% reduction of cervical cancer incidence caused by HPV16 (71.6%, 95% confidence interval (CI) [69.1%, 73.9%]) and a 54% reduction of cancer caused by HPV18 (53.8%, 95% CI [40.6%, 63.1%]). One case of HPV16-caused cervical cancer could be prevented for every 5,527 women attending screening (number needed to screen, NNS). Prevention of one case of HPV16-caused cervical cancer required follow-up of 147 HPV16–positive women (number needed to follow-up, NNF). The NNS and NNF were up to 40 to 500 times higher for HPV types commonly screened for with lower oncogenic potential (HPV35,39,51,56,59,66,68). For women below 30 years of age, NNS and NNF for HPV16 were 4,747 and 289, respectively, but >220,000 and >16,000 for HPV35,39,51,56,59,66,68. All estimates were either age-standarized or age-stratified. The primary limitation of our study is that NNS is dependent on the HPV prevalence that can differ between populations and over time. However, it can readily be recalculated in other settings and monitored when HPV type-specific prevalence changes. Other limitations include that in some age groups, there was little data and extrapolations had to be made. Finally, there were very few cervical cancer cases associated with certain HPV types in young age group. CONCLUSIONS: In this study, we observed that the impact of cervical cancer screening varies depending on the HPV type screened for. Estimating and monitoring the impact of screening by HPV type can facilitate the design of effective and efficient HPV-based cervical screening programs. TRIAL REGISTRATION: ClinicalTrials.gov with numbers NCT00479375, NCT01511328.