Is mammography screening history a predictor of future breast cancer risk?

Abstract

Inspired by the model by Walter and Day for risk of cervical cancer following negative screens, one might hypothesize that women in a mammography screening programme with a certain number of negative screens had a lower remaining breast cancer risk than that of women in general. We studied whether number of negative screens was a predictor for a low remaining breast cancer risk in women participating in the mammography screening programmes in Stockholm, Copenhagen and Funen. Data were collected from the mammography screening programmes in Stockholm, Sweden (1989-2012), Copenhagen, Denmark (1991-2009) and Funen, Denmark (1993-2009), and linked to the respective cancer registries. We calculated cumulative hazard rates for breast cancer in women in cohorts defined by age at entry and number of negative screens for the maximum follow-up period in each screening centre. For all centres and cohorts, the cumulative hazard were parallel for all number of negative screens, from after the time, when the women were scheduled to be invited for the next screen. This means that the remaining breast cancer risk is similar no matter how many negative screens a woman have had. Number of negative screens was not a predictor of a low remaining breast cancer risk in women participating in the mammography screening programmes in Stockholm, Sweden, Copenhagen and Funen, Denmark. The history of previous negative screens is therefore not suitable for personalisation of mammography screening.

Number of prior negative screening outcomes does not influence future risk of breast cancer

“…We used three different approaches to evaluate whether women with a certain number of prior negative screening outcomes could be identified as a low-risk group. We used a cumulative hazard rate of breast cancer as described in Andersen et al [6]. Additionally, we estimated the absolute and relative risk of breast cancer.…”

“…For each age cohort, the cumulative hazard rate of breast cancer was calculated as a function of time since first screen, and by number of negative screening outcomes. As in Andersen et al [6], we used visual inspection to evaluate whether the cumulative hazards from two groups were proportional from a certain point in time, indicating that the two groups had the same breast cancer risk from that point on.…”

“…around 2 years after the start of the curve). As in Andersen et al [6], we focused on the comparison between groups of women with N − 1 negative screening outcomes and N negative outcomes in the time period more than 2 years after 'the start of the curve', to exclude possible screen-detected breast cancers in the following screening round. Sudden jumps in the cumulative hazard curve were typically related to mathematical instabilities as women gradually entered or a large fraction left the group.…”

“…We did not observe any reduction in breast cancer risk among women with high versus low number of negative screening outcomes, in BreastScreen Norway. We used the same study design as Andersen et al [6], and had similar access to complete, individual-level data on screening history and follow-up. Our study confirmed the conclusions in their study, with remarkably similar results.…”

“…Inspired by Walter and Day [5], Andersen et al introduced a hypothesis that women with a certain number of prior negative screening outcomes could be identified as a low risk group [6]. They used data from organized mammographic screening in Sweden and Denmark, and found negative screening outcomes were not a predictor of breast cancer risk, and therefore not suitable for stratified screening.…”