Background The knowledge on HPV type-specific long-term absolute risks (AR) and population attributable proportions (PAR) for CIN2+ is limited. With the SwedeScreen population-based randomised controlled trial the effect of HPV testing in primary screening was evaluated.

Methods Overall 12,527 women (32–38 years) were randomised 1:1 to the intervention (cytology, HPV testing) and control arm (cytology, no action on HPV tests). Registry-based follow-up for cytological and histological test results was done (1997–2011). Type-specific ARs and PARs of CIN2+ were calculated. Poisson regression estimated the relative risk (RR) of new CIN2+. Multivariate analysis adjusted for co-infections. Women were censored at date of first CIN2+ or last registered cytology.

Results Over the entire follow-up, the joint PAR for 14 HR-HPV types was similar in the intervention and control arms (69.3% versus 68.1%). AR, RR and PAR were highest during the first screening round but risks were high throughout follow-up. HR-HPV+ women developed CIN2+: 1–3 years 13.6%, 3–6 years 6.4%, later 4.5%. RRs: 89.5, 37.9, 12.2 and 9.0 during the first, second, third screening rounds and for > 9 years of follow-up. Different HPV types tended to confer different risks over time: HPV18 increased, HPV16 and HPV31 stable, and others decreased. The HR types clustered in a highest, medium and a low AR groups (HPV16/18/31/33: 31–42%, HPV35/45/52/56/58: 13.8–24.8%, HPV39/51/59/66/68: AR < 11%). HPV16 contributed to the greatest proportion of CIN2+ in the population (first round PAR 38.8%), followed by HPV52 (9.6%), HPV31 (7.0%) HPV18 (5.9%) and HPV45 (5.2%).

Conclusion HPV screening had minimal effect on the proportion of CIN2+ lesions caused by the HPV types screened for. HR-HPV-associated risks for CIN2+ continue to be strongly elevated over long-term (9–14 years) follow-up, particularly for HPV16, 18, 31 and particularly for CIN3+ lesions. The seven HR-HPV types 16/18/31/33/45/52/58 cause 73.9% of CIN2+ lesions. All 14 HR types cause 86.9%.