Human papillomavirus-based cervical screening and long-term cervical cancer risk: a randomised health-care policy trial in Sweden

Background: Human papillomavirus (HPV)-based cervical screening is a globally recommended public health policy. Randomised clinical trials find superior performance of primary HPV-based screening compared with cytology for preventing cervical cancer. However, additional evidence from real-world public health policies is needed. In preplanned secondary analysis of a randomised health-care policy trial in Sweden we aimed to evaluate which policy provided better protection against invasive cervical cancer, after two full rounds of screening.

Methods: The organised cervical screening programme in the capital region of Sweden invited all women aged 30-64 years and eligible for screening to a randomised health-care policy trial of HPV-based versus cytology-based screening. During 2014-16, 395 725 eligible women were randomly assigned (non-concealed) to either policy and the invasive cervical cancer incidences over 8 years of follow-up were compared. Intention-to-screen analyses included all invited women and per-protocol analyses the women that attended baseline screening according to protocol. This trial is registered with ClinicalTrials.gov, NCT01511328.

Findings: Women invited to HPV-based cervical screening had a 17% lower invasive cervical cancer risk compared with women invited to cytology (hazard ratio [HR] 0·83, 95% CI 0·70-0·98). Women participating in HPV-based screening had a 28% lower invasive cervical cancer risk compared with women participating with cytology (HR 0·72, 95% CI 0·54-0·95). Adverse events were a higher number of referrals to colposcopy with biopsy in the HPV policy (15 832 [7·5%] of 212 199 in intention to screen analyses and 9968 [9·0%] of 110 176 per protocol at baseline) than in the cytology policy (12 650 [6·9%] of 183 120 in intention to screen analyses, and 7179 [7·9%] of 90 821 per protocol at baseline). Women who were HPV-negative at baseline had invasive cervical cancer risks of 1·3 (95% CI 0·6-2·4) per 100 000 person-years, whereas the risk for women with normal cytology was 9·1 (6·7-11·8) per 100 000 person-years. HPV-positive women with negative cytology triage had invasive cervical cancer risks of 79·2 per 100 000 person-years and HPV 16 or HPV 18-positive women with negative cytology triage had risks of 318·2 per 100 000 person-years.

Interpretation: This randomised policy trial found HPV-based screening to be superior for preventing invasive cervical cancer in the real-world setting. A single baseline HPV-negative test was associated with a very low invasive cervical cancer risk after 8 years. However, HPV positivity with negative cytology triage was associated with high invasive cervical cancer risks.

Funding: Region Stockholm, Swedish Cancer Society, and European Union Horizon 2020.