ARTISTIC: A randomised trial of human papillomavirus (HPV) testing in primary cervical screening

Abstract

Objectives: Primary cervical screening uses cytology to detect cancer precursor lesions [cervical intraepithelial neoplasia stage 3 or beyond (CIN3+)]. Human papillomavirus (HPV) testing could add sensitivity as an adjunct to cytology or as a first test, reserving cytology for HPV-positive women. This study addresses the questions: Does the combination of cytology and HPV testing achieve a reduction in incident CIN3+?; Is HPV testing cost-effective in primary cervical screening?; Is its use associated with adverse psychosocial or psychosexual effects?; and How would it perform as an initial screening test followed by cytology for HPV positivity? Design: ARTISTIC was a randomised trial of cervical cytology versus cervical cytology plus HPV testing, evaluated over two screening rounds, 3 years apart. Round 1 would detect prevalent disease and round 2 a combination of incident and undetected disease from round 1. Setting: Women undergoing routine cervical screening in the NHS programme in Greater Manchester. Participants: In total 24,510 women aged 20–64 years were enrolled between July 2001 and September 2003. Interventions: HPV testing was performed on the liquid-based cytology (LBC) sample obtained at screening. Women were randomised in a ratio of 3:1 to have the HPV test result revealed and acted upon if persistently positive in cytology-negative cases or concealed. A detailed health economic evaluation and a psychosocial and psychosexual assessment were also performed. Main outcome measures: The primary outcome was CIN3+ in round 2. Secondary outcomes included an economic assessment and psychosocial effects. A large HPV genotyping study was also conducted. Results: In round 1 there were 313 CIN3+ lesions, representing a prevalence in the revealed and concealed arms of 1.27% and 1.31% respectively (p = 0.81). Round 2 (30–48 months) involved 14,230 (58.1%) of the women screened in round 1 and only 31 CIN3+ were detected; the CIN3 rate was not significantly different between the revealed and concealed arms. A less restrictive definition of round 2 (26–54 months) increased CIN3+ to 45 and CIN3+ incidence in the arms was significantly different (p = 0.05). There was no difference in CIN3+ between the arms when rounds 1 and 2 were combined. Prevalence of highrisk HPV types was age-dependent. Overall prevalence of HPV16/18 increased with severity of yskaryosis. Mean costs per woman in round 1 were £72 and £56 for the revealed and concealed arms (p < 0.001); an age-adjustment reduced these mean costs to £65 and £52. Incremental cost-effectiveness ratio for detecting additional CIN3+ by adding HPV testing to LBC screening in round 1 was £38,771. Age-adjusted mean cost for LBC primary screening with HPV triage was £39 compared with £48 for HPV primary screening with LBC triage. HPV testing did not appear to cause significant psychosocial distress. Conclusions: Routine HPV testing did not add significantly to the effectiveness of LBC in this study. No significant adverse psychosocial effects were detected. It would not be cost-effective to screen with cytology and HPV combined but HPV testing, as either triage or initial test triaged by cytology, would be cheaper than cytology without HPV testing. LBC would not benefit from combination with HPV; it is highly effective as primary screening but HPV testing has twin advantages of high negative predictive value and automated platforms enabling high throughput. HPV primary screening would require major contraction and reconfiguration of laboratory services. Follow-up continues in ARTISTIC while maintaining concealment for a further 3-year round of screening, which will help in screening protocol development for the post-vaccination era.