Modelling the impact of detecting and treating ductal carcinoma in situ in a breast screening programme

Abstract

Objectives: Screening has substantially increased the detection of carcinoma in situ of the breast (CIS). Opinions vary as to whether this constitutes over-diagnosis or an opportunity to interrupt breast cancer's natural history. In England, incidence of invasive cancer and CIS increased in women of screening age (50–64 years), leading to a subsequent deficit in invasive incidence in women aged 65–69 years immediately beyond the invited age range. We aimed to model underlying incidence of invasive cancer and CIS expected in the absence of screening, and to quantify the likely relative contributions of their early detection to the observed reduction in invasive cancer in women of postscreening age. Setting: UK NHS breast screening programme in England. Methods: Poisson regression modelling was used to establish the underlying incidence of invasive and in situ cancers in the absence of screening. We then estimated age- and year-specific excess detection rates attributable to screening. Applying these to population figures we estimated conservatively the relative contributions of early diagnosis of CIS and invasive cancer at 50–64 years of age to the subsequent deficit in invasive cancer in women beyond invitation age (65–69 years), for screening early in the programme and at steady state. Results: Our model estimated a 1.6% annual increase in incidence, giving an estimated deficit of 4.22 invasive cancers per 10,000 women aged 65–69 years in 1996. Carcinoma in situ contributed 13–17% to the deficit, assuming a mean six year lead time and 75–100% progression to invasive cancer. At steady state, with current screening performance and with lead times of 3–4 years (invasive cancer) and 6–9 years (CIS), invasive incidence might be reduced by 5–6 cancers per 10,000 women aged 65–69 years in 2010 (15–20% of underlying incidence), CIS contributing 20–40%. Discussion: The longer lead time associated with CIS attenuates the impact its early detection has on subsequent invasive incidence. At steady state screening, its detection contributes significantly to the deficit in invasive incidence. Our results suggest that, cancer for cancer, there may be just as much benefit in detecting and treating a case of CIS as there is in treating a case of invasive cancer.