Reducing Overtreatment Associated With Overdiagnosis in Cervical Cancer screening-A Model-Based Benefit-Harm Analysis for Austria

Int J Cancer. 2020 Aug 15;147(4):1131-1142. doi: 10.1002/ijc.32849. Epub 2020 Jan 13.

Abstract

A general concern exists that cervical cancer screening using human papillomavirus (HPV) testing may lead to considerable overtreatment. We evaluated the trade-off between benefits and overtreatment among different screening strategies differing by primary tests (cytology, p16/Ki-67, HPV alone or in combinations), interval, age and diagnostic follow-up algorithms. A Markov state-transition model calibrated to the Austrian epidemiological context was used to predict cervical cancer cases, deaths, overtreatments and incremental harm-benefit ratios (IHBR) for each strategy. When considering the same screening interval, HPV-based screening strategies were more effective compared to cytology or p16/Ki-67 testing (e.g., relative reduction in cervical cancer with biennial screening: 67.7% for HPV + Pap cotesting, 57.3% for cytology and 65.5% for p16/Ki-67), but were associated with increased overtreatment (e.g., 19.8% more conizations with biennial HPV + Papcotesting vs. biennial cytology). The IHBRs measured in unnecessary conizations per additional prevented cancer-related death were 31 (quinquennial Pap + p16/Ki-67-triage), 49 (triennial Pap + p16/Ki-67-triage), 58 (triennial HPV + Pap cotesting), 66 (biennial HPV + Pap cotesting), 189 (annual Pap + p16/Ki-67-triage) and 401 (annual p16/Ki-67 testing alone). The IHBRs increased significantly with increasing screening adherence rates and slightly with lower age at screening initiation, with a reduction in HPV incidence or with lower Pap-test sensitivity. Depending on the accepted IHBR threshold, biennial or triennial HPV-based screening in women as of age 30 and biennial cytology in younger women may be considered in opportunistic screening settings with low or moderate adherence such as in Austria. In organized settings with high screening adherence and in postvaccination settings with lower HPV prevalence, the interval may be prolonged.

Keywords: Papanicolaou; benefit-harm analysis; cervical cancer; decision analysis; human papillomavirus; p16/Ki-67; screening.